Make Codex and Claude Code subscriptions go further: a local model router that gives routine agent work to economical models and keeps your strongest model for the hard parts.
Rust ai-agentsai-gatewayclaude-codecodexcoding-agentcost-optimizationdeepseekllm-gatewayllm-routerlocal-firstmodel-routingrusttask-routingtauri
{
"schema": "hiroute.model-metadata-inference-rules/v1",
"as_of": "2026-09-13",
"policy": "Deterministic closure rules for provider-scoped metadata records. A rule never overrides an observed fact and is always recorded per field as basis=inferred with its rule_key. unsupported and not-applicable are determinate outcomes distinct from unknown. Rules must stay anchored to the cited official sources.",
"semantics_updates": {
"capability_state": [
"conditional",
"not-applicable",
"supported",
"unknown",
"unsupported"
],
"cost_hint_state": [
"not-recorded",
"recorded"
],
"execution_fit_state": [
"native-text-representable",
"not-applicable",
"unsupported"
],
"metadata_completeness_state": [
"complete",
"missing",
"not-applicable",
"partial"
],
"metadata_fact_state": [
"conflict",
"entitlement-dependent",
"known",
"not-applicable",
"runtime-required",
"shared-total-budget",
"unknown"
],
"token_limit_state": [
"conflict",
"entitlement-dependent",
"exact",
"not-applicable",
"runtime-required",
"shared-total-budget",
"unknown"
]
},
"encoding_rules_additions": [
"every provider-scoped metadata field is closed to a determinate outcome: an observed fact, a conservative inference recorded per field as basis=inferred with its rule_key, or an explicit unsupported/not-applicable outcome; indeterminate values do not survive closure",
"an inferred value is never presented as an observed assertion: each inference rule cites official or family evidence, and its per-field application stays inspectable through field_provenance"
],
"model_rules": [
{
"rule_key": "rule.codex.gpt-6.1-sol-output/v1",
"record_kind": "model",
"collected_on": "2026-09-30",
"match": {
"provider_id_pattern": "^openai-codex$",
"upstream_model_id_pattern": "^gpt-6\\.1-sol$"
},
"assignments": {
"max_output_tokens": {
"state": "known",
"value": 128000
}
},
"reason": "The authenticated Codex directory publishes context and effort profiles but no output limit. Conservatively inherit the same GPT-6.1 Sol model's public 128,000 output limit as an explicitly labeled inference, not a measured subscription entitlement. Runtime account eligibility remains separate.",
"evidence_refs": [
"source-175",
"source-177"
]
},
{
"rule_key": "rule.codex.auto-review/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^codex-auto-review$"
},
"assignments": {
"capability_hints.reasoning": "not-applicable",
"capability_hints.streaming": "not-applicable",
"capability_hints.tool": "not-applicable",
"capability_hints.vision": "not-applicable",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "runtime-required",
"value": null
},
"max_output_tokens": {
"state": "runtime-required",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "unsupported",
"reason": "Codex internal auto-review reviewer agent; it is not a user-selectable text model and exposes no first-party chat/Responses contract"
}
},
"reason": "The Codex auto-review mode is an internal reviewer agent, not a selectable model; the repository has no dedicated reviewer-agent protocol, so text-model routing does not apply and no static limits are published.",
"evidence_refs": [
"source-080"
]
},
{
"rule_key": "rule.codex.gpt-5.3-codex-spark/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.3-codex-spark$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Spark's reviewed current executable definition carries 128,000 context / 8,192 output and the xhigh reasoning profile; the Codex subscription listing exposes it as a real-time coding model.",
"evidence_refs": [
"source-079",
"source-080"
]
},
{
"rule_key": "rule.openai.gpt-image-generation/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-image-(1\\.5|2)$"
},
"assignments": {
"capability_hints.reasoning": "not-applicable",
"capability_hints.streaming": "not-applicable",
"capability_hints.tool": "not-applicable",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "not-applicable",
"value": null
},
"max_output_tokens": {
"state": "not-applicable",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "not-applicable",
"reason": "Image generation/editing model served by the images API; the native text Responses execution contract does not apply"
}
},
"reason": "Both models are served only by v1/images/generations and v1/images/edits with image output; they expose no chat/Responses surface, so text token limits and text-routing capabilities are not applicable rather than unknown.",
"evidence_refs": [
"source-081",
"source-082",
"source-083"
]
},
{
"rule_key": "rule.openai.gpt-5.6/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.6-(sol|terra|luna)(-pro)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Official GPT-5.6 model pages document 1,050,000 context, 128,000 max output, text/image input, function calling, streaming, and reasoning.effort none..max for Sol, Terra, and Luna.",
"evidence_refs": [
"source-084",
"source-085",
"source-086"
]
},
{
"rule_key": "rule.openai.gpt-5.5/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.5(-pro)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The official GPT-5.5 model page documents 1,050,000 context, 128,000 max output, text/image input, function calling, streaming, and reasoning.effort none..xhigh.",
"evidence_refs": [
"source-087"
]
},
{
"rule_key": "rule.openai.gpt-5.4-mini-nano/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.4-(mini|nano)$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 400000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The small GPT-5.4 tiers follow the conservative documented GPT-5.x coding surface: 400,000 context and 128,000 max output with the shared effort ladder.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-5.4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.4(-pro)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The GPT-5.4 frontier tier follows the documented GPT-5.x limits of 1,050,000 context and 128,000 max output with image input and the effort ladder.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-6-astra/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-6-astra(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The official GPT-6 Astra model page documents 1,050,000 context, 128,000 max output, text/image input, function calling, streaming, and reasoning.effort low..max.",
"evidence_refs": [
"source-088"
]
},
{
"rule_key": "rule.openai.gpt-5.3-codex/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.3-codex$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 400000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The documented GPT-5.3 Codex surface is 400,000 context, 128,000 max output, text/image input, and reasoning.effort low..xhigh.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-5.1-5.2/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5\\.[12](-codex(-max|-mini)?)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 400000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The GPT-5.1/5.2 generation follows the conservative documented GPT-5 coding surface: 400,000 context, 128,000 max output, image input, and effort reasoning.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-5-base/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-5(-chat|-mini|-nano)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 400000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The original GPT-5 family is documented with 400,000 context and 128,000 max output; the minimal/low/medium/high effort ladder is the conservative subset.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.o-series-mini-preview/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^o(1-mini|1-preview)$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The early o1 mini/preview tiers are text-only 128,000-context reasoning models superseded by the current o-series; lifecycle is recorded as deprecated.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.o-series/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^(o1|o3(-mini)?|o4-mini)$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 200000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The o-series reasoning models are documented with a 200,000-token window, image input, and low/medium/high reasoning effort.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-4.1/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-4\\.1(-mini|-nano)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1047576
},
"max_output_tokens": {
"state": "known",
"value": 32768
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GPT-4.1 is a non-reasoning multimodal model documented with a 1,047,576-token window and 32,768 max output; no reasoning-level rendering exists for it.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-4o/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-4o(-mini|-2024-05-13|-2024-08-06|-2024-11-20)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GPT-4o is a non-reasoning multimodal model documented with a 128,000-token window and 16,384 max output.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.openai.gpt-oss/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gpt-oss(-safeguard)?(-(20b|120b))?(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The open-weight GPT-OSS family is documented with a 131,072-token window, 65,536 max output, text-only input, and low/medium/high reasoning effort.",
"evidence_refs": [
"source-019",
"source-022"
]
},
{
"rule_key": "rule.anthropic.claude-5-line/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^claude-(opus|sonnet|haiku|fable|mythos)-?(5(-1)?|5\\.1)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Claude 5 generation (Opus 5, Sonnet 5, Fable 5.1, Mythos 5) is documented with a 1,000,000-token window, 128,000 max output, image input, tool use, and low..max adaptive effort.",
"evidence_refs": [
"source-058",
"source-060",
"source-061"
]
},
{
"rule_key": "rule.anthropic.claude-4-line/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^claude-(opus|sonnet|haiku)-?4([.-][0-9]{1,2})?(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 200000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Claude 4 generation is documented with a 200,000-token standard window, 64,000 max output, image input, and tool use; extended thinking has no documented nameable effort ladder in the pinned sources.",
"evidence_refs": [
"source-060"
]
},
{
"rule_key": "rule.anthropic.claude-3-line/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^claude-3(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 200000
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "retired",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Claude 3 is the pre-reasoning generation: 200,000-token window, 8,192 max output, image input, no reasoning capability, and legacy lifecycle.",
"evidence_refs": [
"source-060"
]
},
{
"rule_key": "rule.google.gemini-preview/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gemini-.*-preview$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "preview",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Preview Gemini 3.x models are documented with a 1,048,576-token window, 65,536 max output, multimodal input, function calling, and low/medium/high thinking levels.",
"evidence_refs": [
"source-003",
"source-004"
]
},
{
"rule_key": "rule.google.gemini/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gemini-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The current Gemini API line is documented with a 1,048,576-token window, 65,536 max output, multimodal input, function calling, and thinking levels low/medium/high.",
"evidence_refs": [
"source-001",
"source-004"
]
},
{
"rule_key": "rule.google.gemma-4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gemma-?4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Gemma 4 is documented in the permissive open-model line with a 262,144-token window, image input, and tool support; 8,192 is the conservative documented output bound.",
"evidence_refs": [
"source-001"
]
},
{
"rule_key": "rule.google.gemma-3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^gemma-?3(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Gemma 3 is the pre-reasoning open-model generation with a 131,072-token window, image input, and no reasoning ladder.",
"evidence_refs": [
"source-001"
]
},
{
"rule_key": "rule.xai.grok-4.20-reasoning/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-4\\.20-.*(reasoning|multi-agent).*$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 30000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Grok 4.20 reasoning variants are documented with a 1,000,000-token window and 30,000 max output; the non-reasoning variant is covered by its own rule.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.xai.grok-non-reasoning/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-.*non-reasoning$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 30000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The non-reasoning Grok variant exposes no thinking control; limits follow the Grok 4.20 surface.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.xai.grok-4.5-4.6/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-4\\.(5|6)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 500000
},
"max_output_tokens": {
"state": "shared-total-budget",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Grok 4.6 documents a 500,000-token window without a separate textual output cap, so max output stays a shared total budget; reasoning effort is always on.",
"evidence_refs": [
"source-031"
]
},
{
"rule_key": "rule.xai.grok-4.3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-4\\.3(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Grok 4.3 is documented with a 1,000,000-token window and 64,000 max output.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.xai.grok-build-composer/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-(build|composer)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Grok coding variants follow the conservative xAI coding surface: 256,000 context and 64,000 max output with effort control.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.xai.grok-legacy/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^grok-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Unspecified Grok variants fall back to the conservative pre-4.3 xAI surface: 131,072 context and 16,384 max output.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.deepseek.v4-flash/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^deepseek-v4-flash(-vision.*)?$",
"provider_id_pattern": "^(?!alibaba-token-plan$).+"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 384000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The retired direct V4 Flash aliases temporarily route to V4.1 Flash; this rule excludes Alibaba Token Plan, where the same ID serves the original text-only V4 Flash.",
"evidence_refs": [
"source-007",
"source-010"
]
},
{
"rule_key": "rule.deepseek.v4-flash-bailian/v1",
"record_kind": "model",
"match": {
"provider_id_pattern": "^alibaba-token-plan$",
"upstream_model_id_pattern": "^deepseek-v4-flash$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 393216
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Alibaba Cloud Model Studio documents its distinct deepseek-v4-flash deployment as text-only with 1,000,000 context and 393,216 maximum output; Token Plan exposes this model ID.",
"evidence_refs": [
"source-137",
"source-153"
]
},
{
"rule_key": "rule.deepseek.v4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^deepseek-v4-pro(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 384000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "DeepSeek V4 Pro is documented with 1,048,576 context and 384,000 max output, text-only input, and a thinking toggle.",
"evidence_refs": [
"source-007",
"source-010"
]
},
{
"rule_key": "rule.deepseek.v4-generic/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^deepseek-v4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 384000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The V4 generation shares the documented 1,048,576 context and 384,000 max output surface with a thinking toggle.",
"evidence_refs": [
"source-007",
"source-010"
]
},
{
"rule_key": "rule.deepseek.v3.2/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^deepseek-v3\\.2(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 163840
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "DeepSeek V3.2 is a superseded text-only generation with a 163,840-token window and 65,536 max output.",
"evidence_refs": [
"source-007"
]
},
{
"rule_key": "rule.deepseek.v3-r1/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^deepseek-(v3|r1)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The earlier DeepSeek V3/R1 line is text-only with a 131,072-token window and 8,192 max output.",
"evidence_refs": [
"source-007"
]
},
{
"rule_key": "rule.qwen.3.8/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3\\.8(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"xhigh"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen 3.8 Max is documented with a 1,000,000-token window, 131,072 max output, multimodal input, and reasoning_effort low/medium/xhigh.",
"evidence_refs": [
"source-041",
"source-042"
]
},
{
"rule_key": "rule.qwen.3.7-plus/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3\\.7-plus$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen 3.7 Plus follows the documented 1,000,000-token / 65,536-output Qwen3 API surface with image input.",
"evidence_refs": [
"source-041"
]
},
{
"rule_key": "rule.qwen.3.7/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3\\.7(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen 3.7 Max is documented with a 1,000,000-token window and 65,536 max output; the text-only variant keeps text input.",
"evidence_refs": [
"source-041"
]
},
{
"rule_key": "rule.qwen.3.6/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3\\.6(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1000000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen 3.6 follows the documented 1,000,000-token / 65,536-output Qwen3 API surface with multimodal input.",
"evidence_refs": [
"source-041"
]
},
{
"rule_key": "rule.qwen.3.5/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3\\.5(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen 3.5 is documented with a 262,144-token window and 8,192 max output; thinking is a model toggle rather than a named ladder.",
"evidence_refs": [
"source-041"
]
},
{
"rule_key": "rule.qwen.coder-next-conflict/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3-coder-next$"
},
"assignments": {
"capability_hints.reasoning": "conditional",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Bailian documents conflicting thinking support for qwen3-coder-next; the conflict is preserved as a determinate conditional capability instead of a guessed boolean.",
"evidence_refs": [
"source-039",
"source-043"
]
},
{
"rule_key": "rule.qwen.coder/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3-coder(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Qwen3 Coder line is a text-only coding surface with a 262,144-token window and 65,536 max output.",
"evidence_refs": [
"source-039",
"source-041"
]
},
{
"rule_key": "rule.qwen3.vl/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3-vl(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Qwen3-VL variants accept image input with the conservative Qwen3 open-model window.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.qwen3.thinking/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3-.*thinking.*$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Qwen3 thinking open models support reasoning with the conservative 131,072-token window.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.qwen3.legacy/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^qwen3(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The earlier Qwen3 open models are text-only with a conservative 131,072-token window.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.zai.glm-5.3-flash/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-5\\.3-flash$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GLM-5.3 Flash is documented with a 1,000,000-token window, 131,072 max output, and image input (the Flash variant carries vision).",
"evidence_refs": [
"source-035",
"source-036"
]
},
{
"rule_key": "rule.zai.glm-5.3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-5\\.3$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GLM-5.3 is documented with a 1,000,000-token window, 131,072 max output, forced thinking with low/high/max, and text input.",
"evidence_refs": [
"source-035",
"source-036"
]
},
{
"rule_key": "rule.zai.glm-5.2/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-5\\.2(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GLM-5.2 is documented with a 1,000,000-token window and 131,072 max output; the maintained evidence lists high/max thinking.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.zai.glm-5.1/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-5\\.1(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 200000
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GLM-5.1 is the superseded previous generation with a 200,000-token window and conservative 16,384 max output.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.zai.glm-5-base/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-?5(-turbo)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 202752
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The base GLM-5 line is documented with a 202,752-token window and conservative 16,384 max output.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.zai.glm-4.x/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-4\\.(5|6|7)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 202752
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The GLM-4.5/4.6/4.7 line is documented with a 202,752-token window, reasoning effort control, and text input.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.zai.glm-4-legacy/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The earlier GLM-4 line is text-only with a 131,072-token window and 8,192 max output.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.zai.glm-5v/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^glm-5v(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 200000
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "GLM-5V Turbo is the vision variant with a 200,000-token window and 131,072 max output.",
"evidence_refs": [
"source-027"
]
},
{
"rule_key": "rule.kimi.k3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^k3(-256k)?$|^kimi-k3$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "entitlement-dependent",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Kimi K3 is documented with a 1,048,576-token window; the completion cap is entitlement-dependent per plan, and always-thinking effort is low/high/max.",
"evidence_refs": [
"source-062",
"source-074"
]
},
{
"rule_key": "rule.kimi.k3-256k/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^k3-256k$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"high",
"max"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The k3-256k code-plan cap serves the same model with a fixed 262,144-token window.",
"evidence_refs": [
"source-075"
]
},
{
"rule_key": "rule.kimi.k2.7-code/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^kimi-k2\\.7-code(-highspeed)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 262144
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Kimi K2.7 Code is documented with a 262,144-token shared window.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.kimi.k2.6/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^kimi-k2\\.6(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Kimi K2.6 is documented with a 256,000-window and 65,536 conservative output bound; 32,768 is only its default, not its maximum.",
"evidence_refs": [
"source-064"
]
},
{
"rule_key": "rule.kimi.k2.5/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^kimi-k2[.-]5(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "shared-total-budget",
"value": null
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Kimi K2.5 shares its 262,144-token total budget between input and output; the generation is superseded by K2.6/K2.7.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.kimi.k2-thinking/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^kimi-k2-thinking(-turbo)?$|^moonshotai/kimi-k2-thinking$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Kimi K2 Thinking is a thinking-mode K2 with a 262,144-token window and conservative 16,384 output bound.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.kimi.k2-legacy/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^kimi-k2(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 16384
},
"lifecycle": "deprecated",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Legacy K2 preview/turbo ids are superseded text-only ids with a 262,144-token window.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.minimax.m3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^minimax-m3$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 512000
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "MiniMax M3 is observed with a 512,000-token configured window; without a first-party model card the conservative output bound is kept.",
"evidence_refs": [
"source-065"
]
},
{
"rule_key": "rule.minimax.m2.7/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^minimax-m2\\.7(-highspeed)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 204800
},
"max_output_tokens": {
"state": "known",
"value": 204800
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "MiniMax M2.7 is documented with a 204,800-token window and Anthropic-compatible output up to 204,800.",
"evidence_refs": [
"source-066",
"source-067"
]
},
{
"rule_key": "rule.minimax.m2/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^minimax-m2(\\.1|\\.5)?(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 196608
},
"max_output_tokens": {
"state": "known",
"value": 32768
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The M2.x line is documented with a 196,608-token window and conservative 32,768 output bound.",
"evidence_refs": [
"source-065"
]
},
{
"rule_key": "rule.tencent.hy4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^hy4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1024000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "preview",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Hunyuan Hy4 preview is documented with a 1,024,000-token window and 64,000 max output.",
"evidence_refs": [
"source-016"
]
},
{
"rule_key": "rule.tencent.hy3/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^hy3(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Hunyuan Hy3 is documented with a 256,000-token window, 192,000 input / 128,000 output, and none/low/high reasoning.",
"evidence_refs": [
"source-015",
"source-016"
]
},
{
"rule_key": "rule.bytedance.seed/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^(doubao-|dola-|seed-).*$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Doubao Seed 2.x line is observed with a 256,000-token window and 128,000 max output; thinking is a toggle.",
"evidence_refs": [
"source-076"
]
},
{
"rule_key": "rule.baidu.ernie-5.1/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^ernie-5\\.1(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "ERNIE 5.1 is observed as a text-only non-reasoning record with a 128,000-token window and 65,536 max output.",
"evidence_refs": [
"source-089"
]
},
{
"rule_key": "rule.baidu.ernie/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^ernie-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "ERNIE 5.x is observed with a 128,000-token window, 65,536 max output, and multimodal input.",
"evidence_refs": [
"source-089"
]
},
{
"rule_key": "rule.xiaomi.mimo-omni/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^mimo-.*(omni|2\\.5)(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "MiMo V2.5 / Omni records are observed with a 1,048,576-token window and 131,072 max output with image input.",
"evidence_refs": [
"source-090"
]
},
{
"rule_key": "rule.xiaomi.mimo/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^mimo-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The MiMo V2 code line is observed with a 1,048,576-token window and text-first input.",
"evidence_refs": [
"source-090"
]
},
{
"rule_key": "rule.stepfun.step-3.7/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^step-3\\.7(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 262144
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Step 3.7 Flash is observed with a 262,144-token window and shared output budget with image input.",
"evidence_refs": [
"source-091"
]
},
{
"rule_key": "rule.stepfun.step/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^step-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 65536
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Step 3.5 is observed with a 262,144-token window and 65,536 max output.",
"evidence_refs": [
"source-091"
]
},
{
"rule_key": "rule.amazon.nova/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^nova-"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 5000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Amazon Nova keeps a conservative lower bound of 128,000 context and 5,000 output from the Nova product documentation; exact per-model limits remain the Bedrock model card's authority.",
"evidence_refs": [
"source-092"
]
},
{
"rule_key": "rule.meta.muse-spark/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^muse-spark-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Meta Muse Spark is observed with a 1,048,576-token window and 131,072 max output across model and contributor tiers.",
"evidence_refs": [
"source-093"
]
},
{
"rule_key": "rule.meta.llama-4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^llama-?4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Llama 4 hosted variants keep the conservative hosted service bound of 131,072 context and 8,192 output.",
"evidence_refs": [
"source-094"
]
},
{
"rule_key": "rule.meta.llama/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^llama-?"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Llama 3.x hosted ids are text-only with a 131,072-token window and 8,192 max output.",
"evidence_refs": [
"source-094"
]
},
{
"rule_key": "rule.nous.hermes-4/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^hermes-4(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Nous Hermes 4 is a Llama-3.1-based reasoning fine-tune with a 131,072-token window and text-only input.",
"evidence_refs": [
"source-095"
]
},
{
"rule_key": "rule.nous.hermes/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^hermes-"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Nous Hermes 3 is a non-reasoning Llama-based fine-tune with a 131,072-token window.",
"evidence_refs": [
"source-095"
]
},
{
"rule_key": "rule.mistral.small/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^mistral-small(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "shared-total-budget",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Mistral Small 4 shares one 256,000-token total budget between input and output with reasoning effort support.",
"evidence_refs": [
"source-030"
]
},
{
"rule_key": "rule.mistral.large/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^mistral-large(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "shared-total-budget",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Mistral Large 3 documents a 256,000-token window with a shared output budget and multimodal input.",
"evidence_refs": [
"source-029"
]
},
{
"rule_key": "rule.mistral.medium/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^mistral-medium(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Mistral Medium 3.x follows the 262,144-token platform window with a conservative 8,192 output bound.",
"evidence_refs": [
"source-028"
]
},
{
"rule_key": "rule.nvidia.nemotron-omni/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^nemotron-.*omni.*$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Nemotron 3 Nano Omni keeps the 262,144-token Nemotron family window and adds image input.",
"evidence_refs": [
"source-096"
]
},
{
"rule_key": "rule.nvidia.nemotron/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^nemotron-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 262144
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high"
],
"thinking_level_maps": []
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The Nemotron 3 line is text-first with a 262,144-token conservative window and 8,192 output bound and reasoning support.",
"evidence_refs": [
"source-096"
]
},
{
"rule_key": "rule.nvidia.llama-nemotron/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^llama-.*nemotron.*$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Llama-Nemotron derivatives keep the Llama 3.1 hosted bound of 131,072 context and 8,192 output.",
"evidence_refs": [
"source-096"
]
},
{
"rule_key": "rule.cohere.north-mini-code/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^north-mini-code(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Cohere North Mini Code is observed with a 256,000-token window and 64,000 max output.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.cohere.command-a-reasoning/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^command-a-reasoning(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 32000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Command A Reasoning is observed with a 256,000-token window and 32,000 max output.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.cohere.command-a-vision/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^command-a-vision(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "unsupported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 8000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Command A Vision is observed with a 128,000-token window and 8,000 max output; it exposes no tool calling.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.cohere.command-a-plus/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^command-a-plus(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 64000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Command A Plus is observed with a 128,000-token window, 64,000 max output, and multimodal input.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.cohere.command/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^command-"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 256000
},
"max_output_tokens": {
"state": "known",
"value": 8000
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "The original Command A is observed with a 256,000-token window and 8,000 max output.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.upstage.solar/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^solar-"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Upstage Solar keeps the conservative 131,072-token window and 8,192 output bound from its platform documentation.",
"evidence_refs": [
"source-098"
]
},
{
"rule_key": "rule.longcat.longcat/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^longcat-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 1048576
},
"max_output_tokens": {
"state": "known",
"value": 131072
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "LongCat 2.0 is observed with a 1,048,576-token window and 131,072 max output.",
"evidence_refs": [
"source-099"
]
},
{
"rule_key": "rule.venice.uncensored/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^venice-uncensored(-.*)?$"
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "supported",
"input_modalities": [
"image",
"text"
],
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Venice Uncensored is observed with a 128,000-token window and 8,192 output bound; no reasoning ladder is documented.",
"evidence_refs": [
"source-100"
]
},
{
"rule_key": "rule.poolside.laguna/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^laguna-"
},
"assignments": {
"capability_hints.reasoning": "supported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "Poolside Laguna code models keep the conservative 131,072-token window and 8,192 output bound.",
"evidence_refs": [
"source-101"
]
},
{
"rule_key": "rule.client.alias-default/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "^default$"
},
"assignments": {
"capability_hints.reasoning": "not-applicable",
"capability_hints.streaming": "not-applicable",
"capability_hints.tool": "not-applicable",
"capability_hints.vision": "not-applicable",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "not-applicable",
"value": null
},
"max_output_tokens": {
"state": "not-applicable",
"value": null
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "not-applicable",
"reason": "Client-side alias literal, not a callable upstream model identity"
}
},
"reason": "The literal id `default` is a client alias, not a model: no text-routing capabilities or limits apply.",
"evidence_refs": [
"source-079"
]
},
{
"rule_key": "rule.fallback.unclassified-text-model/v1",
"record_kind": "model",
"match": {
"upstream_model_id_pattern": "."
},
"assignments": {
"capability_hints.reasoning": "unsupported",
"capability_hints.streaming": "supported",
"capability_hints.tool": "supported",
"capability_hints.vision": "unsupported",
"input_modalities": [
"text"
],
"context_tokens": {
"state": "known",
"value": 131072
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"lifecycle": "active",
"reasoning_rendering_hints": {
"state": "not-applicable"
},
"execution_fit": {
"state": "native-text-representable"
}
},
"reason": "No family-specific official model card matched this provider-scoped id. Applied the conservative project default for an unclassified text response model: tools and streaming supported, vision and reasoning unsupported, smallest documented family-free limits (131,072/8,192), lifecycle active unless observed otherwise. Replace with exact evidence when the provider publishes a model card.",
"evidence_refs": [
"source-079"
]
},
{
"rule_key": "rule.codex.gpt6-model-context/v1",
"record_kind": "model",
"collected_on": "2026-10-08",
"match": {
"provider_id_pattern": "^openai-codex$",
"upstream_model_id_pattern": "^(gpt-6-(astra|sol|luna)|gpt-6\\.1-sol)$"
},
"assignments": {
"context_tokens": {
"state": "known",
"value": 1050000
}
},
"reason": "Use the exact public model's documented 1,050,000-token ceiling for the maintained Codex subscription model mapping. This is explicitly inferred for the subscription, not a measured account entitlement. Retain the native client directory's default/max-window settings as raw source assertions; do not substitute its 272,000 default for the model maximum. Account inventory and any independently proven endpoint limit remain separate execution constraints. No API token price is assigned to a subscription.",
"evidence_refs": [
"source-088",
"source-126",
"source-127",
"source-175"
]
}
],
"provider_rules": [
{
"rule_key": "rule.provider.auth.external-process/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(claude-cli|copilot-acp)$"
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"external_process"
]
},
"reason": "CLI/ACP-backed providers authenticate through the local process rather than a stored API key.",
"evidence_refs": [
"source-102"
]
},
{
"rule_key": "rule.provider.auth.vertex/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^vertex$"
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"vertex"
]
},
"reason": "Vertex AI uses Google Cloud application default credentials.",
"evidence_refs": [
"source-056"
]
},
{
"rule_key": "rule.provider.auth.bedrock/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^bedrock$"
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"aws_sdk"
]
},
"reason": "Amazon Bedrock uses AWS SDK credentials.",
"evidence_refs": [
"source-055"
]
},
{
"rule_key": "rule.provider.auth.copilot/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(copilot|github-copilot|copilot-model-aliases)$"
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"copilot"
]
},
"reason": "The Copilot integrations authenticate with a GitHub Copilot token.",
"evidence_refs": [
"source-103"
]
},
{
"rule_key": "rule.provider.auth.oauth/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(gemini|qwen-oauth|minimax-oauth)$"
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"oauth_external"
]
},
"reason": "These providers expose a documented external OAuth login beside API keys.",
"evidence_refs": [
"source-104",
"source-105",
"source-106"
]
},
{
"rule_key": "rule.provider.auth.not-applicable/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(custom|preferred-silent-default|moa)$"
},
"assignments": {
"domain": "authentication",
"completeness": "not-applicable"
},
"reason": "These entries are local placeholders or aggregations; credential requirements are defined by the user or composed at runtime, so no fixed authentication contract exists.",
"evidence_refs": [
"source-079"
]
},
{
"rule_key": "rule.provider.auth.api-key/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "."
},
"assignments": {
"domain": "authentication",
"authentication_candidates": [
"api_key"
]
},
"reason": "Every remaining provider is a documented API service that authenticates bearer API keys on its HTTP interface.",
"evidence_refs": [
"source-022",
"source-053",
"source-107"
]
},
{
"rule_key": "rule.provider.discovery.openai-chat/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^openai-chat$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.openai.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for openai-chat.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.provider.discovery.openai-api/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^openai-api$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.openai.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for openai-api.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.provider.discovery.openai/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^openai$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.openai.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for openai.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.provider.discovery.moonshot/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^moonshot$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.moonshot.cn/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for moonshot.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.provider.discovery.kimi/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kimi$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.moonshot.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for kimi.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.provider.discovery.kimi-coding/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kimi-coding$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.kimi.com/coding/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for kimi-coding.",
"evidence_refs": [
"source-075"
]
},
{
"rule_key": "rule.provider.discovery.kimi-coding-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kimi-coding-cn$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.kimi.com/coding/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for kimi-coding-cn.",
"evidence_refs": [
"source-075"
]
},
{
"rule_key": "rule.provider.discovery.anthropic/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^anthropic$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.anthropic.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for anthropic.",
"evidence_refs": [
"source-060"
]
},
{
"rule_key": "rule.provider.discovery.tencent/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^tencent$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.lkeap.cloud.tencent.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for tencent.",
"evidence_refs": [
"source-015"
]
},
{
"rule_key": "rule.provider.discovery.tencent-tokenhub/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^tencent-tokenhub$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.lkeap.cloud.tencent.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for tencent-tokenhub.",
"evidence_refs": [
"source-015"
]
},
{
"rule_key": "rule.provider.discovery.tencent-tokenplan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^tencent-tokenplan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.lkeap.cloud.tencent.com/plan/v3/models"
]
},
"reason": "The provider documents an authenticated remote model directory for tencent-tokenplan.",
"evidence_refs": [
"source-017"
]
},
{
"rule_key": "rule.provider.discovery.xai/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.x.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xai.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.discovery.xai-curated-extras/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-curated-extras$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.x.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xai-curated-extras.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.discovery.xai-static-fallback/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-static-fallback$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.x.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xai-static-fallback.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.discovery.xai-top/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-top$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.x.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xai-top.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.discovery.google/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^google$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://generativelanguage.googleapis.com/v1beta/openai/models"
]
},
"reason": "The provider documents an authenticated remote model directory for google.",
"evidence_refs": [
"source-005"
]
},
{
"rule_key": "rule.provider.discovery.gemini/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^gemini$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://generativelanguage.googleapis.com/v1beta/openai/models"
]
},
"reason": "The provider documents an authenticated remote model directory for gemini.",
"evidence_refs": [
"source-005"
]
},
{
"rule_key": "rule.provider.discovery.qwen/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^qwen$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://dashscope.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for qwen.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.provider.discovery.qwen-token-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^qwen-token-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for qwen-token-plan.",
"evidence_refs": [
"source-044"
]
},
{
"rule_key": "rule.provider.discovery.zai/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^zai$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.z.ai/api/paas/v4/models"
]
},
"reason": "The provider documents an authenticated remote model directory for zai.",
"evidence_refs": [
"source-035"
]
},
{
"rule_key": "rule.provider.discovery.zhipu/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^zhipu$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://open.bigmodel.cn/api/paas/v4/models"
]
},
"reason": "The provider documents an authenticated remote model directory for zhipu.",
"evidence_refs": [
"source-026"
]
},
{
"rule_key": "rule.provider.discovery.deepseek/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^deepseek$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.deepseek.com/models"
]
},
"reason": "The provider documents an authenticated remote model directory for deepseek.",
"evidence_refs": [
"source-007"
]
},
{
"rule_key": "rule.provider.discovery.mistral/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^mistral$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.mistral.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for mistral.",
"evidence_refs": [
"source-028"
]
},
{
"rule_key": "rule.provider.discovery.groq/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^groq$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.groq.com/openai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for groq.",
"evidence_refs": [
"source-020"
]
},
{
"rule_key": "rule.provider.discovery.openrouter/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^openrouter$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://openrouter.ai/api/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for openrouter.",
"evidence_refs": [
"source-053"
]
},
{
"rule_key": "rule.provider.discovery.nvidia/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^nvidia$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://integrate.api.nvidia.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for nvidia.",
"evidence_refs": [
"source-096"
]
},
{
"rule_key": "rule.provider.discovery.venice/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^venice$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.venice.ai/api/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for venice.",
"evidence_refs": [
"source-100"
]
},
{
"rule_key": "rule.provider.discovery.ollama-cloud/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^ollama-cloud$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ollama.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for ollama-cloud.",
"evidence_refs": [
"source-108"
]
},
{
"rule_key": "rule.provider.discovery.together/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^together$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.together.xyz/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for together.",
"evidence_refs": [
"source-109"
]
},
{
"rule_key": "rule.provider.discovery.cerebras/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^cerebras$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.cerebras.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for cerebras.",
"evidence_refs": [
"source-110"
]
},
{
"rule_key": "rule.provider.discovery.featherless/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^featherless$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.featherless.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for featherless.",
"evidence_refs": [
"source-111"
]
},
{
"rule_key": "rule.provider.discovery.chutes/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^chutes$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://llm.chutes.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for chutes.",
"evidence_refs": [
"source-112"
]
},
{
"rule_key": "rule.provider.discovery.deepinfra/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^deepinfra$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.deepinfra.com/v1/openai/models"
]
},
"reason": "The provider documents an authenticated remote model directory for deepinfra.",
"evidence_refs": [
"source-113"
]
},
{
"rule_key": "rule.provider.discovery.novita/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^novita$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.novita.ai/v3/openai/models"
]
},
"reason": "The provider documents an authenticated remote model directory for novita.",
"evidence_refs": [
"source-114"
]
},
{
"rule_key": "rule.provider.discovery.stepfun/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^stepfun$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.stepfun.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for stepfun.",
"evidence_refs": [
"source-091"
]
},
{
"rule_key": "rule.provider.discovery.stepfun-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^stepfun-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.stepfun.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for stepfun-plan.",
"evidence_refs": [
"source-091"
]
},
{
"rule_key": "rule.provider.discovery.xiaomi/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xiaomi$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.mimo.xiaomi.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xiaomi.",
"evidence_refs": [
"source-090"
]
},
{
"rule_key": "rule.provider.discovery.xiaomi-token-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xiaomi-token-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.mimo.xiaomi.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for xiaomi-token-plan.",
"evidence_refs": [
"source-090"
]
},
{
"rule_key": "rule.provider.discovery.qianfan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^qianfan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://qianfan.baidubce.com/v2/models"
]
},
"reason": "The provider documents an authenticated remote model directory for qianfan.",
"evidence_refs": [
"source-089"
]
},
{
"rule_key": "rule.provider.discovery.longcat/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^longcat$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.longcat.chat/openai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for longcat.",
"evidence_refs": [
"source-099"
]
},
{
"rule_key": "rule.provider.discovery.volcengine/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^volcengine$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ark.cn-beijing.volces.com/api/v3/models"
]
},
"reason": "The provider documents an authenticated remote model directory for volcengine.",
"evidence_refs": [
"source-076"
]
},
{
"rule_key": "rule.provider.discovery.volcengine-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^volcengine-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ark.cn-beijing.volces.com/api/v3/models"
]
},
"reason": "The provider documents an authenticated remote model directory for volcengine-plan.",
"evidence_refs": [
"source-076"
]
},
{
"rule_key": "rule.provider.discovery.byteplus/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^byteplus$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ark.ap-southeast.bytepluses.com/api/v3/models"
]
},
"reason": "The provider documents an authenticated remote model directory for byteplus.",
"evidence_refs": [
"source-115"
]
},
{
"rule_key": "rule.provider.discovery.byteplus-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^byteplus-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ark.ap-southeast.bytepluses.com/api/v3/models"
]
},
"reason": "The provider documents an authenticated remote model directory for byteplus-plan.",
"evidence_refs": [
"source-115"
]
},
{
"rule_key": "rule.provider.discovery.huggingface/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^huggingface$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://router.huggingface.co/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for huggingface.",
"evidence_refs": [
"source-116"
]
},
{
"rule_key": "rule.provider.discovery.baseten/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^baseten$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://inference.baseten.co/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for baseten.",
"evidence_refs": [
"source-117"
]
},
{
"rule_key": "rule.provider.discovery.fireworks/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^fireworks$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.fireworks.ai/inference/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for fireworks.",
"evidence_refs": [
"source-118"
]
},
{
"rule_key": "rule.provider.discovery.gmi/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^gmi$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.gmi-serving.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for gmi.",
"evidence_refs": [
"source-119"
]
},
{
"rule_key": "rule.provider.discovery.kilocode/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kilocode$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://kilocode.ai/api/openrouter/models"
]
},
"reason": "The provider documents an authenticated remote model directory for kilocode.",
"evidence_refs": [
"source-120"
]
},
{
"rule_key": "rule.provider.discovery.opencode-zen/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^opencode-zen$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://opencode.ai/zen/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for opencode-zen.",
"evidence_refs": [
"source-121"
]
},
{
"rule_key": "rule.provider.discovery.opencode-go/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^opencode-go$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://opencode.ai/zen/go/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for opencode-go.",
"evidence_refs": [
"source-121"
]
},
{
"rule_key": "rule.provider.discovery.not-applicable/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(claude-cli|copilot-acp|custom|preferred-silent-default|moa|bedrock|vertex|azure-foundry|router|ai-gateway|copilot-model-aliases|openai-codex)$"
},
"assignments": {
"domain": "discovery",
"completeness": "not-applicable"
},
"reason": "These providers expose no authenticated HTTP model directory; discovery is a local process, deployment, or console fact instead.",
"evidence_refs": [
"source-049",
"source-102"
]
},
{
"rule_key": "rule.provider.endpoint.openai/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(openai-api|openai-chat|openai)$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.openai.com/v1"
],
"protocol_candidates": [
"responses",
"chat_completions"
],
"completeness": "partial"
},
"reason": "OpenAI Platform documents the shared /v1 base for the Responses and Chat Completions interfaces.",
"evidence_refs": [
"source-022"
]
},
{
"rule_key": "rule.provider.endpoint.not-applicable/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^(claude-cli|copilot-acp|custom|preferred-silent-default|moa|copilot-model-aliases)$"
},
"assignments": {
"domain": "endpoint",
"completeness": "not-applicable"
},
"reason": "These entries expose no first-party HTTP endpoint in this dataset; requests are mediated by a local process, a user-provided address, or an alias catalog.",
"evidence_refs": [
"source-102"
]
},
{
"rule_key": "rule.provider.identity.fallback-id/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "."
},
"assignments": {
"domain": "identity",
"completeness": "complete"
},
"reason": "The provider exposes exactly one unambiguous identity string: a single asserted name or the pinned scanner's provider id.",
"evidence_refs": [
"source-079"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://dashscope-intl.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba-cn$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://dashscope.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba-cn.",
"evidence_refs": [
"source-039"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba-coding-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba-coding-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://coding-intl.dashscope.aliyuncs.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba-coding-plan.",
"evidence_refs": [
"source-043"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba-coding-plan-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba-coding-plan-cn$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://coding.dashscope.aliyuncs.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba-coding-plan-cn.",
"evidence_refs": [
"source-043"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba-token-plan/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba-token-plan$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba-token-plan.",
"evidence_refs": [
"source-044"
]
},
{
"rule_key": "rule.provider.discovery2.alibaba-token-plan-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^alibaba-token-plan-cn$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for alibaba-token-plan-cn.",
"evidence_refs": [
"source-044"
]
},
{
"rule_key": "rule.provider.discovery2.arcee/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^arcee$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.arcee.ai/api/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for arcee.",
"evidence_refs": [
"source-122"
]
},
{
"rule_key": "rule.provider.discovery2.minimax/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^minimax$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.minimax.io/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for minimax.",
"evidence_refs": [
"source-065"
]
},
{
"rule_key": "rule.provider.discovery2.minimax-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^minimax-cn$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.minimaxi.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for minimax-cn.",
"evidence_refs": [
"source-071"
]
},
{
"rule_key": "rule.provider.discovery2.minimax-oauth/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^minimax-oauth$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.minimax.io/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for minimax-oauth.",
"evidence_refs": [
"source-065"
]
},
{
"rule_key": "rule.provider.discovery2.nous/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^nous$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://inference-api.nousresearch.com/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for nous.",
"evidence_refs": [
"source-095"
]
},
{
"rule_key": "rule.provider.discovery2.upstage/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^upstage$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.upstage.ai/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for upstage.",
"evidence_refs": [
"source-098"
]
},
{
"rule_key": "rule.provider.discovery2.opencode-free/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^opencode-free$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://opencode.ai/zen/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for opencode-free.",
"evidence_refs": [
"source-121"
]
},
{
"rule_key": "rule.provider.discovery2.cohere/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^cohere$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.cohere.com/v2/models"
]
},
"reason": "The provider documents an authenticated remote model directory for cohere.",
"evidence_refs": [
"source-097"
]
},
{
"rule_key": "rule.provider.discovery2.github-copilot/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^github-copilot$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.individual.githubcopilot.com/models"
]
},
"reason": "The provider documents an authenticated remote model directory for github-copilot.",
"evidence_refs": [
"source-103"
]
},
{
"rule_key": "rule.provider.discovery2.copilot/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^copilot$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.githubcopilot.com/models"
]
},
"reason": "The provider documents an authenticated remote model directory for copilot.",
"evidence_refs": [
"source-103"
]
},
{
"rule_key": "rule.provider.discovery2.vercel-ai-gateway/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^vercel-ai-gateway$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://ai-gateway.vercel.sh/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for vercel-ai-gateway.",
"evidence_refs": [
"source-124"
]
},
{
"rule_key": "rule.provider.discovery2.meta/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^meta(-ai)?$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://api.meta.ai/v1/models"
]
},
"reason": "Meta documents an authenticated remote model directory on api.meta.ai for its Model API surface, covering the aliased meta and meta-ai provider profiles.",
"evidence_refs": [
"source-093"
]
},
{
"rule_key": "rule.provider.discovery2.opencode/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^opencode$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
],
"models_url_candidates": [
"https://opencode.ai/zen/v1/models"
]
},
"reason": "The provider documents an authenticated remote model directory for opencode.",
"evidence_refs": [
"source-121"
]
},
{
"rule_key": "rule.provider.discovery.nebius-modes/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^nebius-token-factory$"
},
"assignments": {
"domain": "discovery",
"discovery_modes": [
"remote_models"
]
},
"reason": "The Nebius Token Factory directory URL is already recorded; the modes marker adds the remote-inventory semantics.",
"evidence_refs": [
"source-123"
]
},
{
"rule_key": "rule.provider.endpoint2.kimi-coding/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kimi-coding$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.kimi.com/coding/v1"
],
"protocol_candidates": [
"chat_completions",
"messages"
],
"completeness": "partial"
},
"reason": "The provider documents this endpoint surface for kimi-coding.",
"evidence_refs": [
"source-075"
]
},
{
"rule_key": "rule.provider.endpoint2.kimi-coding-cn/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^kimi-coding-cn$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.kimi.com/coding/v1"
],
"protocol_candidates": [
"chat_completions",
"messages"
],
"completeness": "partial"
},
"reason": "The provider documents this endpoint surface for kimi-coding-cn.",
"evidence_refs": [
"source-075"
]
},
{
"rule_key": "rule.provider.endpoint2.moonshot/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^moonshot$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.moonshot.cn/v1"
],
"protocol_candidates": [
"chat_completions"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for moonshot.",
"evidence_refs": [
"source-063"
]
},
{
"rule_key": "rule.provider.endpoint2.tencent/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^tencent$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.lkeap.cloud.tencent.com/v1"
],
"protocol_candidates": [
"chat_completions"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for tencent.",
"evidence_refs": [
"source-015"
]
},
{
"rule_key": "rule.provider.endpoint2.xai-curated-extras/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-curated-extras$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.x.ai/v1"
],
"protocol_candidates": [
"responses"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for xai-curated-extras.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.endpoint2.xai-static-fallback/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-static-fallback$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.x.ai/v1"
],
"protocol_candidates": [
"responses"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for xai-static-fallback.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.endpoint2.xai-top/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^xai-top$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.x.ai/v1"
],
"protocol_candidates": [
"responses"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for xai-top.",
"evidence_refs": [
"source-033"
]
},
{
"rule_key": "rule.provider.endpoint2.vercel-ai-gateway/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^vercel-ai-gateway$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://ai-gateway.vercel.sh/v1"
],
"protocol_candidates": [
"chat_completions"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for vercel-ai-gateway.",
"evidence_refs": [
"source-124"
]
},
{
"rule_key": "rule.provider.endpoint2.meta-ai/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^meta-ai$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://api.meta.ai/v1"
],
"protocol_candidates": [
"responses"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for meta-ai.",
"evidence_refs": [
"source-093"
]
},
{
"rule_key": "rule.provider.endpoint2.ollama-cloud/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^ollama-cloud$"
},
"assignments": {
"domain": "endpoint",
"base_url_candidates": [
"https://ollama.com"
],
"protocol_candidates": [
"chat_completions"
],
"completeness": "complete"
},
"reason": "The provider documents this endpoint surface for ollama-cloud.",
"evidence_refs": [
"source-108"
]
},
{
"rule_key": "rule.provider.discovery.qwen-oauth/v1",
"record_kind": "provider",
"match": {
"provider_id_pattern": "^qwen-oauth$"
},
"assignments": {
"domain": "discovery",
"completeness": "not-applicable"
},
"reason": "The pinned scan records no authenticated models-directory contract for the Qwen OAuth consumer surface.",
"evidence_refs": [
"source-105"
]
}
],
"evidence_source_additions": [
{
"source_key": "source-079",
"authority": "learn.chatgpt.com",
"source_kind": "provider-documentation",
"locator": "https://learn.chatgpt.com/docs/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-080",
"authority": "learn.chatgpt.com",
"source_kind": "provider-documentation",
"locator": "https://learn.chatgpt.com/docs/sandboxing/auto-review",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-081",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-image-1.5",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-082",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-image-2",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-083",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/guides/image-generation",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-084",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-5.6-luna",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-085",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-5.6-sol",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-086",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-5.6-terra",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-087",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-5.5",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-088",
"authority": "developers.openai.com",
"source_kind": "provider-documentation",
"locator": "https://developers.openai.com/api/docs/models/gpt-6-astra.md",
"collected_on": "2026-10-08",
"bytes_digest": "sha256:b7c19b3bb89e87a392a0f3f8b732c5e7c9f0bf558cbe3022c491575a947fcef8",
"digest_state": "captured"
},
{
"source_key": "source-089",
"authority": "cloud.baidu.com",
"source_kind": "provider-documentation",
"locator": "https://cloud.baidu.com/doc/qianfan-api/index.html",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-090",
"authority": "mimo.xiaomi.com",
"source_kind": "provider-documentation",
"locator": "https://mimo.xiaomi.com/docs/en-US/updates/model",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-091",
"authority": "platform.stepfun.com",
"source_kind": "provider-documentation",
"locator": "https://platform.stepfun.com/docs/overview/model-overview",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-092",
"authority": "docs.aws.amazon.com",
"source_kind": "provider-documentation",
"locator": "https://docs.aws.amazon.com/nova/latest/userguide/what-is-nova.html",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-093",
"authority": "ai.developer.meta.com",
"source_kind": "provider-documentation",
"locator": "https://ai.developer.meta.com/docs/models/",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-094",
"authority": "docs.aws.amazon.com",
"source_kind": "provider-documentation",
"locator": "https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-meta-llama.html",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-095",
"authority": "nousresearch.com",
"source_kind": "provider-documentation",
"locator": "https://nousresearch.com/hermes4/",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-096",
"authority": "build.nvidia.com",
"source_kind": "provider-documentation",
"locator": "https://build.nvidia.com/nvidia/nemotron-3-super-120b-a12b/modelcard",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-097",
"authority": "docs.cohere.com",
"source_kind": "provider-documentation",
"locator": "https://docs.cohere.com/docs/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-098",
"authority": "console.upstage.ai",
"source_kind": "provider-documentation",
"locator": "https://console.upstage.ai/docs/capabilities/chat",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-099",
"authority": "longcat.chat",
"source_kind": "provider-documentation",
"locator": "https://longcat.chat/platform/docs/zh/APIDocs.html",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-100",
"authority": "docs.venice.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.venice.ai/api-reference/endpoint/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-101",
"authority": "docs.poolside.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.poolside.ai/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-102",
"authority": "docs.anthropic.com",
"source_kind": "provider-documentation",
"locator": "https://docs.anthropic.com/en/docs/claude-code/overview",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-103",
"authority": "docs.github.com",
"source_kind": "provider-documentation",
"locator": "https://docs.github.com/en/copilot",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-104",
"authority": "ai.google.dev",
"source_kind": "provider-documentation",
"locator": "https://ai.google.dev/gemini-api/docs/oauth",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-105",
"authority": "help.aliyun.com",
"source_kind": "provider-documentation",
"locator": "https://help.aliyun.com/en/model-studio/oauth",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-106",
"authority": "platform.minimax.io",
"source_kind": "provider-documentation",
"locator": "https://platform.minimax.io/docs/api-reference/api-overview",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-107",
"authority": "openrouter.ai",
"source_kind": "provider-documentation",
"locator": "https://openrouter.ai/docs/api-reference/authentication",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-108",
"authority": "docs.ollama.com",
"source_kind": "provider-documentation",
"locator": "https://docs.ollama.com/cloud",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-109",
"authority": "docs.together.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.together.ai/reference/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-110",
"authority": "inference-docs.cerebras.ai",
"source_kind": "provider-documentation",
"locator": "https://inference-docs.cerebras.ai/api-reference/models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-111",
"authority": "featherless.ai",
"source_kind": "provider-documentation",
"locator": "https://featherless.ai/docs",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-112",
"authority": "docs.chutes.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.chutes.ai/",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-113",
"authority": "deepinfra.com",
"source_kind": "provider-documentation",
"locator": "https://deepinfra.com/docs/api/get-models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-114",
"authority": "novita.ai",
"source_kind": "provider-documentation",
"locator": "https://novita.ai/docs/api-reference/model-apis-llm-list-models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-115",
"authority": "docs.byteplus.com",
"source_kind": "provider-documentation",
"locator": "https://docs.byteplus.com/en/docs/ModelArk/ListModels",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-116",
"authority": "huggingface.co",
"source_kind": "provider-documentation",
"locator": "https://huggingface.co/docs/inference-providers/index",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-117",
"authority": "docs.baseten.co",
"source_kind": "provider-documentation",
"locator": "https://docs.baseten.co/inference/model-apis/overview",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-118",
"authority": "docs.fireworks.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.fireworks.ai/api-reference/list-models",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-119",
"authority": "docs.gmicloud.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.gmicloud.ai/",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-120",
"authority": "kilocode.ai",
"source_kind": "provider-documentation",
"locator": "https://kilocode.ai/docs",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-121",
"authority": "opencode.ai",
"source_kind": "provider-documentation",
"locator": "https://opencode.ai/docs/zen",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-122",
"authority": "docs.arcee.ai",
"source_kind": "provider-documentation",
"locator": "https://docs.arcee.ai/",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-123",
"authority": "docs.nebius.com",
"source_kind": "provider-documentation",
"locator": "https://docs.nebius.com/studio/inference/api",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
},
{
"source_key": "source-124",
"authority": "vercel.com",
"source_kind": "provider-documentation",
"locator": "https://vercel.com/docs/ai-gateway",
"collected_on": "2026-09-13",
"bytes_digest": null,
"digest_state": "not-captured"
}
],
"record_additions": [
{
"model_record_key": "hermes-agent/openai-codex/codex-auto-review",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "codex-auto-review",
"display_name": "Codex Auto Review",
"context_tokens": {
"state": "runtime-required",
"value": null
},
"max_output_tokens": {
"state": "runtime-required",
"value": null
},
"input_modalities": [
"text"
],
"capability_hints": {
"reasoning": "not-applicable",
"streaming": "not-applicable",
"tool": "not-applicable",
"vision": "not-applicable"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-080"
}
],
"source_locations": [
{
"source_key": "source-080"
}
],
"execution_fit": {
"state": "unsupported",
"reason": "Codex internal auto-review reviewer agent; it is not a user-selectable text model and exposes no first-party chat/Responses contract"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "partial",
"modalities": "complete",
"reasoning_rendering": "not-applicable"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-5.3-codex-spark",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-5.3-codex-spark",
"display_name": "GPT-5.3 Codex Spark",
"context_tokens": {
"state": "known",
"value": 128000
},
"max_output_tokens": {
"state": "known",
"value": 8192
},
"input_modalities": [
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "unsupported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"xhigh"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-079"
}
],
"source_locations": [
{
"source_key": "source-079"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [
"gpt-5-3-codex"
],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-5.5",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-5.5",
"display_name": "GPT-5.5",
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-087"
}
],
"source_locations": [
{
"source_key": "source-087"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-5.6-luna",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-5.6-luna",
"display_name": "GPT-5.6 Luna",
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-084"
}
],
"source_locations": [
{
"source_key": "source-084"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-5.6-sol",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-5.6-sol",
"display_name": "GPT-5.6 Sol",
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-085"
}
],
"source_locations": [
{
"source_key": "source-085"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [
"gpt-5-6-sol"
],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-5.6-terra",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-5.6-terra",
"display_name": "GPT-5.6 Terra",
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-086"
}
],
"source_locations": [
{
"source_key": "source-086"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-6-astra",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-6-astra",
"display_name": "GPT-6 Astra",
"context_tokens": {
"state": "known",
"value": 1050000
},
"max_output_tokens": {
"state": "known",
"value": 128000
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "supported",
"streaming": "supported",
"tool": "supported",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [
"low",
"medium",
"high",
"xhigh",
"max"
],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-088"
}
],
"source_locations": [
{
"source_key": "source-088"
}
],
"execution_fit": {
"state": "native-text-representable"
},
"normalized_model_matches": [
"gpt-6-astra"
],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "complete"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-image-1.5",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-image-1.5",
"display_name": "GPT Image 1.5",
"context_tokens": {
"state": "not-applicable",
"value": null
},
"max_output_tokens": {
"state": "not-applicable",
"value": null
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "not-applicable",
"streaming": "not-applicable",
"tool": "not-applicable",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-081"
}
],
"source_locations": [
{
"source_key": "source-081"
}
],
"execution_fit": {
"state": "not-applicable",
"reason": "Image generation/editing model served by the images API; the native text Responses execution contract does not apply"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "not-applicable"
},
"usable_for": []
},
{
"model_record_key": "hermes-agent/openai-codex/gpt-image-2",
"provider_record_key": "hermes-agent/openai-codex",
"upstream_model_id": "gpt-image-2",
"display_name": "GPT Image 2",
"context_tokens": {
"state": "not-applicable",
"value": null
},
"max_output_tokens": {
"state": "not-applicable",
"value": null
},
"input_modalities": [
"image",
"text"
],
"capability_hints": {
"reasoning": "not-applicable",
"streaming": "not-applicable",
"tool": "not-applicable",
"vision": "supported"
},
"reasoning_rendering_hints": {
"reasoning_effort_maps": [],
"supported_reasoning_efforts": [],
"thinking_level_maps": []
},
"lifecycle": "active",
"provenance_refs": [
{
"kind": "evidence-source",
"source_key": "source-082"
}
],
"source_locations": [
{
"source_key": "source-082"
}
],
"execution_fit": {
"state": "not-applicable",
"reason": "Image generation/editing model served by the images API; the native text Responses execution contract does not apply"
},
"normalized_model_matches": [],
"metadata_completeness": {
"capabilities": "complete",
"cost": "missing",
"identity": "complete",
"lifecycle": "complete",
"limits": "complete",
"modalities": "complete",
"reasoning_rendering": "not-applicable"
},
"usable_for": []
}
]
}