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.native-rating-maintenance/v1",
"source_bundle": "assets/model-data/current/rating-seed.json",
"source_bytes_digest": "sha256:2f3b02526a9263e756c3a94e4d74336aca36724abd984fbec4a0a89833619b9f",
"metadata_catalog": "assets/model-data/current/inputs/metadata-catalog.json",
"metadata_catalog_bytes_digest": "sha256:9b69fb47d9a0a87e3c5ef75446a4a6ad36890828841bb40a05c158a3546532ed",
"client_discovery_runs": "assets/model-data/current/inputs/client-discovery-runs.json",
"client_discovery_runs_bytes_digest": "sha256:7412c6cc46c90fe0f11611e9dc28873bd316fb77a67a0fa2ee782a7b6ecbeb67",
"snapshot_version": "hiroute-20260913/current-1",
"method_revision": "preserved-explicit-native/v1",
"evidence": {
"model.groq.openai-gpt-oss-20b": {
"id": "preserved-seed/groq-medium",
"identity_url": "https://console.groq.com/docs/model/openai/gpt-oss-20b",
"collected_on": "2026-09-06",
"numeric_source": "pinned source bundle data.ratings; not a score measured by the linked official page",
"native_identity_source": "pinned source bundle rating_snapshot metadata and native_reasoning",
"harness_task_scope": "preserved seed; original common harness unavailable",
"raw_unit": "HiRoute star tenths, unchanged",
"reason": "Preserve only the explicitly identified configuration as estimated. No per-dimension score inferred; no renormalization."
},
"model.modelscope.qwen3-coder-30b-a3b-instruct": {
"id": "preserved-seed/qwen-non-thinking",
"identity_url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct",
"collected_on": "2026-09-06",
"numeric_source": "pinned source bundle data.ratings; not a score measured by the linked official page",
"native_identity_source": "pinned source bundle rating_snapshot metadata and native_reasoning",
"harness_task_scope": "preserved seed; original common harness unavailable",
"raw_unit": "HiRoute star tenths, unchanged",
"reason": "Preserve only the explicitly identified configuration as estimated. No per-dimension score inferred; no renormalization."
},
"model.openai.gpt-5.3-codex-spark": {
"id": "preserved-seed/spark-xhigh",
"identity_url": "https://openai.com/index/introducing-gpt-5-3-codex-spark/",
"collected_on": "2026-09-06",
"numeric_source": "pinned source bundle data.ratings; not a score measured by the linked official page",
"native_identity_source": "pinned source bundle rating_snapshot metadata and native_reasoning",
"harness_task_scope": "preserved seed; original common harness unavailable",
"raw_unit": "HiRoute star tenths, unchanged",
"reason": "Preserve only the explicitly identified configuration as estimated. No per-dimension score inferred; no renormalization."
}
},
"pending_families": [],
"reference_models": [
{
"model": {
"model_configuration_id": "model.anthropic.claude-opus-5",
"revision": 1,
"display_name": "Claude Opus 5",
"publisher_id": "publisher.anthropic",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1000000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "output_config.effort",
"profiles": [
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"native_render_convention": "claude_adaptive_effort_messages",
"sources": [
"https://platform.claude.com/docs/en/models/opus-5/overview",
"https://platform.claude.com/docs/en/build-with-claude/effort"
],
"collected_on": "2026-09-06",
"identity_note": "Official Claude model ID claude-opus-5. Every effort binds explicit thinking.type=adaptive under confirmed QUESTION-40001; disabled-thinking combinations are outside this snapshot, not merged scores. No endpoint/Offer or execution eligibility is added.",
"rating_method": "Unknown: no matching adaptive-mode measurement on the retained common scale. No legacy default scores reused."
},
{
"model": {
"model_configuration_id": "model.anthropic.claude-sonnet-5",
"revision": 1,
"display_name": "Claude Sonnet 5",
"publisher_id": "publisher.anthropic",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1000000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "output_config.effort",
"profiles": [
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"native_render_convention": "claude_adaptive_effort_messages",
"sources": [
"https://platform.claude.com/docs/en/models/sonnet-5/overview",
"https://platform.claude.com/docs/en/build-with-claude/effort"
],
"collected_on": "2026-09-06",
"identity_note": "Official Claude model ID claude-sonnet-5. Every effort binds explicit thinking.type=adaptive under confirmed QUESTION-40001; disabled-thinking combinations are outside this snapshot, not merged scores. No endpoint/Offer or execution eligibility is added.",
"rating_method": "Unknown: no matching adaptive-mode measurement on the retained common scale. No legacy default scores reused."
},
{
"model": {
"model_configuration_id": "model.bailian.qwen3-max-2026-01-23",
"revision": 1,
"display_name": "Qwen3 Max 2026-01-23",
"publisher_id": "publisher.alibaba-qwen",
"capabilities": {
"tool": true,
"vision": false,
"streaming": true,
"context_tokens": 262144,
"max_output_tokens": 65536
}
},
"native_capability": {
"kind": "toggle",
"parameter": "enable_thinking"
},
"sources": [
"https://help.aliyun.com/zh/model-studio/list-models",
"https://help.aliyun.com/zh/model-studio/model-qwen3-max",
"https://help.aliyun.com/zh/model-studio/compatibility-of-openai-with-dashscope"
],
"collected_on": "2026-09-10",
"identity_note": "Official immutable qwen3-max-2026-01-23 identity for the China (Beijing) Bailian PAYG OpenAI-compatible endpoint. The authenticated /api/v1/models directory remains the runtime eligibility authority; this entry alone does not prove an account can call the model.",
"rating_method": "Unknown: no comparable measurement on the retained common scale. No score is inferred from provider capability or pricing documentation.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.bailian.qwen3-max-2026-01-23.payg.chat",
"revision": 1,
"model_configuration_id": "model.bailian.qwen3-max-2026-01-23",
"connector_id": "connector.bailian.p0",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.bailian.payg.cn.v1",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.bailian.payg.cn.v1.chat",
"upstream_protocol": "chat_completions",
"upstream_model_id": "qwen3-max-2026-01-23",
"required_adapter_ref": "adapter.openai-chat.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:e390f2e13258cdd5b7b8605566fba44567ed24545ca7011b999b80f4da532aab"
}
],
"offers": [
{
"offer_id": "offer.bailian.payg.cn",
"revision": 1,
"endpoint_profile_id": "endpoint.bailian.payg.cn.v1",
"endpoint_profile_revision": 1,
"service_offering_id": "model-studio",
"entitlement_id": "payg-api-key",
"usage_scope": "workspace",
"region_id": "cn",
"model_configuration_ids": [
"model.bailian.qwen3-max-2026-01-23"
],
"billing_class": "paid",
"evidence_digest": "sha256:e390f2e13258cdd5b7b8605566fba44567ed24545ca7011b999b80f4da532aab"
}
],
"price_rates": []
}
},
{
"model": {
"model_configuration_id": "model.deepseek.v4-pro-0813",
"revision": 1,
"display_name": "DeepSeek V4 Pro 0813",
"publisher_id": "publisher.deepseek",
"capabilities": {
"tool": true,
"vision": false,
"streaming": true,
"context_tokens": 1000000,
"max_output_tokens": 384000
}
},
"native_capability": {
"kind": "fixed",
"profile": "provider-default"
},
"sources": [
"https://api-docs.deepseek.com/quick_start/pricing/",
"https://api-docs.deepseek.com/api/create-response/"
],
"collected_on": "2026-09-06",
"identity_note": "Official model version DeepSeek-V4-Pro-0813 is distinct from rolling API alias deepseek-v4-pro. The maintained catalog now records a toggle without an exact portable render path, so runtime execution keeps provider-default reasoning and does not retain the older selectable effort projection.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization."
},
{
"model": {
"model_configuration_id": "model.moonshot.kimi-k2.5",
"revision": 1,
"display_name": "Kimi K2.5",
"publisher_id": "publisher.moonshot",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 262144,
"max_output_tokens": 262144
}
},
"native_capability": {
"kind": "toggle",
"parameter": "thinking.type"
},
"sources": [
"https://github.com/MoonshotAI/Kimi-K2.5",
"https://platform.kimi.com/docs/guide/use-kimi-vision-model",
"https://www.kimi.ai/help/kimi-api/api-troubleshooting"
],
"collected_on": "2026-09-06",
"identity_note": "Official K2.5 model version. Maximum output shares the 256*1024 total context budget and is reduced by prompt tokens; 32768 is a default, not its maximum. No hosted endpoint readiness is inferred.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization."
},
{
"model": {
"model_configuration_id": "model.openai.gpt-5.5",
"revision": 1,
"display_name": "GPT-5.5",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-5.5"
],
"collected_on": "2026-09-13",
"identity_note": "Official model ID gpt-5.5. Promoted onto the client-bundled Codex subscription Responses endpoint only after the closed provider-scoped metadata record (known 1,050,000/128,000 limits, tool/streaming/image input, reasoning.effort none..xhigh) matched the promoted definition. No dated snapshot and no endpoint association beyond the bundled Codex profile.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-5.5.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-5.5",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-5.5",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-5.5"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-5.6-luna",
"revision": 1,
"display_name": "GPT-5.6 Luna",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-5.6-luna"
],
"collected_on": "2026-09-13",
"identity_note": "Official model ID gpt-5.6-luna. Promoted onto the client-bundled Codex subscription Responses endpoint only after the closed provider-scoped metadata record (known 1,050,000/128,000 limits, tool/streaming/image input, reasoning.effort none..max) matched the promoted definition.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-5.6-luna.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-5.6-luna",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-5.6-luna",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-5.6-luna"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-5.6-sol",
"revision": 1,
"display_name": "GPT-5.6 Sol",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-5.6-sol"
],
"collected_on": "2026-09-13",
"identity_note": "Official model ID gpt-5.6-sol. Promoted onto the client-bundled Codex subscription Responses endpoint only after the closed provider-scoped metadata record matched the promoted definition. The OpenAI Platform identity remains a separate product binding.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-5.6-sol.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-5.6-sol",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-5.6-sol",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-5.6-sol"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-5.6-terra",
"revision": 1,
"display_name": "GPT-5.6 Terra",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-5.6-terra"
],
"collected_on": "2026-09-13",
"identity_note": "Official model ID gpt-5.6-terra. Promoted onto the client-bundled Codex subscription Responses endpoint only after the closed provider-scoped metadata record (known 1,050,000/128,000 limits, tool/streaming/image input, reasoning.effort none..max) matched the promoted definition.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-5.6-terra.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-5.6-terra",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-5.6-terra",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-5.6-terra"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-6-astra",
"revision": 1,
"display_name": "GPT-6 Astra",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-6-astra"
],
"collected_on": "2026-09-06",
"identity_note": "Official model ID gpt-6-astra. No invented dated snapshot; catalog revision records collected reference facts. Promoted onto the client-bundled Codex subscription Responses endpoint only after the closed provider-scoped metadata record (known 1,050,000/128,000 limits, tool/streaming/image input, reasoning.effort low..max) matched this definition.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-6-astra.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-6-astra",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-6-astra",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-6-astra"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-6-luna",
"revision": 1,
"display_name": "GPT-6 Luna",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-6-luna.md"
],
"collected_on": "2026-10-08",
"identity_note": "Exact GPT-6 Luna identity with the existing Codex Responses endpoint and subscription offer; current CPA inventory must observe gpt-6-luna. The 1,050,000 model context ceiling is a labeled public-model inference for this subscription, not an account entitlement measurement. Preserve native none/low/medium/high/xhigh/max profiles, image/tool/streaming facts and the documented 128,000 output bound. Do not inherit API prices.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-6-luna.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-6-luna",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-6-luna",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-6-luna"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-6-sol",
"revision": 1,
"display_name": "GPT-6 Sol",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"none",
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://developers.openai.com/api/docs/models/gpt-6-sol"
],
"collected_on": "2026-09-29",
"identity_note": "Official model ID gpt-6-sol. The client-bundled Codex subscription Responses projection is gated by the current account's observed CPA inventory; the official model facts establish 1,050,000/128,000 limits, text/image input, tools, streaming, and reasoning.effort none..max, not account availability.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-6-sol.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-6-sol",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-6-sol",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-6-sol"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"model": {
"model_configuration_id": "model.openai.gpt-6.1-sol",
"revision": 1,
"display_name": "GPT-6.1 Sol",
"publisher_id": "publisher.openai",
"capabilities": {
"tool": true,
"vision": true,
"streaming": true,
"context_tokens": 1050000,
"max_output_tokens": 128000
}
},
"native_capability": {
"kind": "discrete",
"parameter": "reasoning_effort",
"profiles": [
"low",
"medium",
"high",
"xhigh",
"max"
]
},
"sources": [
"https://chatgpt.com/backend-api/codex/models?client_version=0.159.0",
"https://developers.openai.com/api/docs/models/gpt-6.1-sol.md"
],
"collected_on": "2026-10-08",
"identity_note": "Exact GPT-6.1 Sol model ceiling is 1,050,000, projected to the Codex subscription as a labeled public-model inference. Preserve the captured native default/max-window settings in source assertions instead of using the default as the model maximum. Low/medium/high/xhigh/max are native reasoning.effort profiles; ultra automatic delegation is not projected. Current account availability requires observed CPA inventory and no subscription API token price is inferred.",
"rating_method": "Unknown: no comparable measurement on the retained legacy scale. No new normalization.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.openai.gpt-6.1-sol.codex.responses",
"revision": 1,
"model_configuration_id": "model.openai.gpt-6.1-sol",
"connector_id": "connector.cpa.codex",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.cpa.codex.responses",
"upstream_protocol": "responses",
"upstream_model_id": "gpt-6.1-sol",
"required_adapter_ref": "adapter.openai-responses.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
],
"offers": [
{
"offer_id": "offer.codex.subscription",
"revision": 1,
"endpoint_profile_id": "endpoint.cpa.codex",
"endpoint_profile_revision": 1,
"service_offering_id": "codex-subscription",
"entitlement_id": "codex-subscription",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.openai.gpt-6.1-sol"
],
"billing_class": "subscription",
"evidence_digest": "sha256:0000000000000000000000000000000000000000000000000000000000000000"
}
]
}
},
{
"source_model_key": "deepseek-v4-1-flash",
"model_configuration_id": "model.deepseek.v4-1-flash",
"publisher_id": "publisher.deepseek",
"native_capability": {
"kind": "fixed",
"profile": "provider-default"
},
"sources": [
"https://api-docs.deepseek.com/updates/"
],
"collected_on": "2026-10-19",
"identity_note": "Directory intent for the official deepseek-flash rolling alias. The October 17 announcement does not identify the version executed by another hosting product. Account availability and the saved connection still require the normal registered-connection checks.",
"catalog_intent": {
"endpoint_capabilities": [
{
"capability_id": "cap.deepseek.v4-1-flash.official.chat",
"revision": 1,
"model_configuration_id": "model.deepseek.v4-1-flash",
"connector_id": "connector.deepseek.p0",
"connector_revision": 1,
"endpoint_profile_id": "endpoint.deepseek.official.global.v1",
"endpoint_profile_revision": 1,
"protocol_endpoint_id": "endpoint.deepseek.official.global.v1.chat",
"upstream_protocol": "chat_completions",
"upstream_model_id": "deepseek-flash",
"required_adapter_ref": "adapter.openai-chat.v1",
"required_adapter_revision": 1,
"evidence_digest": "sha256:7912d8c2c263cf0b9beea39d6085db85d84d437e6038d57a30356c2184143b74"
}
],
"offers": [
{
"offer_id": "offer.deepseek.v4-1-flash.official",
"revision": 1,
"endpoint_profile_id": "endpoint.deepseek.official.global.v1",
"endpoint_profile_revision": 1,
"service_offering_id": "official-api",
"entitlement_id": "api-key",
"usage_scope": "account",
"region_id": "global",
"model_configuration_ids": [
"model.deepseek.v4-1-flash"
],
"billing_class": "paid",
"evidence_digest": "sha256:7912d8c2c263cf0b9beea39d6085db85d84d437e6038d57a30356c2184143b74"
}
]
}
}
],
"current_provider_limits": [
{
"model_configuration_id": "model.openrouter.liquid-lfm-2.5-2.6b-free",
"model_record_key": "hermes-agent/openrouter/liquid%2Flfm-2.5-2.6b%3Afree"
},
{
"model_configuration_id": "model.openrouter.nvidia-nemotron-3-super-120b-a12b-free",
"model_record_key": "hermes-agent/openrouter/nvidia%2Fnemotron-3-super-120b-a12b%3Afree"
}
]
}