Skip to content

Self-improving AI Employees for everyone

config: refresh model defaults and restrict global lists to 1M-context

trohitg62c5dbaf1f97
diff --git a/server/config/llm_defaults.json b/server/config/llm_defaults.json
index 2332a6f5..130b1edb 100644
--- a/server/config/llm_defaults.json
+++ b/server/config/llm_defaults.json
@@ -1,7 +1,7 @@
{
- "version": "2026.07.18",
- "last_updated": "2026-07-18",
- "_description": "LLM provider defaults. popular_models drives the ordered global selector and offline fallback independently of token constraints. Connection config drives native SDK and LangChain paths; model specs are enriched by model_registry.json (OpenRouter cache).",
+ "version": "2026.07.23",
+ "last_updated": "2026-07-23",
+ "_description": "LLM provider defaults. popular_models drives the ordered global selector and offline fallback independently of token constraints. POLICY (2026-07-23): popular_models lists only models with >= 1M-token context windows — providers whose models are all below 1M carry an empty list and contribute nothing to the global selector (their models stay reachable via the live per-provider fetch_models dropdown). Connection config drives native SDK and LangChain paths; model specs are enriched by model_registry.json (OpenRouter cache).",
"agent": {
"_description": "Agent execution defaults. recursion_limit and compaction.ratio are sourced from the environment (AGENT_RECURSION_LIMIT, COMPACTION_RATIO via core.config.Settings) — these JSON values are last-resort fallbacks when the env vars aren't loaded (e.g. one-off CLI scripts that bypass Settings). default_temperature stays JSON-resident since it has no env override yet.",
@@ -53,7 +53,7 @@
},
"anthropic": {
"default_model": "claude-opus-4-8",
- "popular_models": ["claude-fable-5", "claude-sonnet-5", "claude-opus-4-8", "claude-opus-4-7", "claude-sonnet-4-6", "claude-haiku-4-5"],
+ "popular_models": ["claude-fable-5", "claude-sonnet-5", "claude-opus-4-8", "claude-opus-4-7", "claude-sonnet-4-6"],
"detection_patterns": ["claude", "anthropic"],
"base_url": "https://api.anthropic.com/v1",
"models_endpoint": "https://api.anthropic.com/v1/models",
@@ -82,7 +82,7 @@
},
"gemini": {
"default_model": "gemini-3.5-flash",
- "popular_models": ["gemini-3.5-flash", "gemini-3.1-pro-preview", "gemini-3.1-flash-lite", "gemini-3-flash-preview", "gemini-2.5-pro", "gemini-2.5-flash", "gemini-2.5-flash-lite", "gemma-4-31b-it", "gemma-4-26b-a4b-it"],
+ "popular_models": ["gemini-3.5-flash", "gemini-3.1-pro-preview", "gemini-3.1-flash-lite", "gemini-3-flash-preview", "gemini-2.5-pro", "gemini-2.5-flash", "gemini-2.5-flash-lite"],
"detection_patterns": ["gemini", "google", "gemma"],
"_detection_patterns_note": "'gemma' makes is_model_valid_for_provider accept the API-served Gemma models (gemma-4-*-it) on gemini nodes — without it the four agent-path call sites silently swap gemma to default_model. Cerebras keeps its own 'gemma-4' pattern for its gemma-4-31b (no -it suffix); is_model_valid_for_provider is per-provider so both coexist. Only the runtime-unused detect_provider_from_model classifier prefers gemini for gemma-* names (dict order).",
"base_url": "https://generativelanguage.googleapis.com/v1beta",
@@ -124,7 +124,7 @@
},
"openrouter": {
"default_model": "anthropic/claude-sonnet-5",
- "popular_models": ["anthropic/claude-sonnet-5", "openai/gpt-5.6-terra", "openai/gpt-5.6-sol", "google/gemini-3.5-flash", "moonshotai/kimi-k3", "x-ai/grok-4.20", "openrouter/auto-beta"],
+ "popular_models": ["anthropic/claude-sonnet-5", "openai/gpt-5.6-terra", "openai/gpt-5.6-sol", "google/gemini-3.5-flash", "moonshotai/kimi-k3", "x-ai/grok-4.20"],
"detection_patterns": ["openrouter"],
"base_url": "https://openrouter.ai/api/v1",
"models_endpoint": "https://openrouter.ai/api/v1/models",
@@ -140,13 +140,15 @@
"temperature_range": [0.0, 2.0]
},
"groq": {
- "default_model": "llama-3.3-70b-versatile",
- "popular_models": ["openai/gpt-oss-120b", "openai/gpt-oss-20b", "llama-3.3-70b-versatile", "llama-3.1-8b-instant", "qwen/qwen3-32b"],
+ "default_model": "openai/gpt-oss-120b",
+ "popular_models": [],
+ "_popular_note": "Empty per the 1M-context policy — no Groq-served model reaches 1M (all top out at 131072). Models remain reachable via live fetch_models.",
"detection_patterns": ["groq"],
"base_url": "https://api.groq.com/openai/v1",
"models_endpoint": "https://api.groq.com/openai/v1/models",
"api_key_param": "api_key",
"max_tokens_param": "max_tokens",
+ "_models_note": "Default moved to openai/gpt-oss-120b (production route). Groq announced deprecation of llama-3.3-70b-versatile and llama-3.1-8b-instant on 2026-06-17 (console.groq.com/docs/deprecations); kept in the list while still served — they fall back to _default constraints after removal.",
"max_output_tokens": {
"openai/gpt-oss-120b": 65536,
"openai/gpt-oss-20b": 65536,
@@ -169,7 +171,8 @@
},
"cerebras": {
"default_model": "gpt-oss-120b",
- "popular_models": ["gpt-oss-120b", "zai-glm-4.7", "gemma-4-31b"],
+ "popular_models": [],
+ "_popular_note": "Empty per the 1M-context policy — no Cerebras model reaches 1M (all 131072). Models remain reachable via live fetch_models.",
"detection_patterns": ["cerebras", "gpt-oss", "zai-glm", "gemma-4"],
"base_url": "https://api.cerebras.ai/v1",
"models_endpoint": "https://api.cerebras.ai/v1/models",
@@ -194,13 +197,16 @@
},
"xai": {
"default_model": "grok-4.20",
- "popular_models": ["grok-4.20", "grok-4.20-multi-agent", "grok-4.5", "grok-4.3", "grok-3"],
+ "popular_models": ["grok-4.20", "grok-4.20-multi-agent", "grok-4.3"],
+ "_popular_note": "grok-4.5 (flagship, released 2026-07-08) and grok-3 excluded per the 1M-context policy (500K / 131072); reachable via live fetch_models. Default stays grok-4.20 (2M) so the default is selectable from the global list.",
"detection_patterns": ["grok", "xai"],
"base_url": "https://api.x.ai/v1",
"models_endpoint": "https://api.x.ai/v1/models",
"api_key_param": "api_key",
"max_tokens_param": "max_tokens",
+ "_models_note": "grok-4.5 (released 2026-07-08): 500K context, configurable reasoning effort (default high), $2/$6 per MTok. Output cap undocumented by xAI — curated 131072 matches the rest of the grok family; registry cache wins when present.",
"max_output_tokens": {
+ "grok-4.5": 131072,
"grok-4.20": 131072,
"grok-4.3": 131072,
"grok-3": 131072,
@@ -218,7 +224,8 @@
},
"deepseek": {
"default_model": "deepseek-v4-flash",
- "popular_models": ["deepseek-v4-flash", "deepseek-v4-pro", "deepseek-chat", "deepseek-reasoner"],
+ "popular_models": ["deepseek-v4-flash", "deepseek-v4-pro"],
+ "_popular_note": "deepseek-chat / deepseek-reasoner dropped from the global list per the 1M-context policy (131072); reachable via live fetch_models.",
"detection_patterns": ["deepseek"],
"base_url": "https://api.deepseek.com",
"models_endpoint": "https://api.deepseek.com/models",
@@ -243,8 +250,10 @@
"temperature_range": [0.0, 2.0]
},
"kimi": {
- "default_model": "kimi-k2.6",
- "popular_models": ["kimi-k3", "kimi-k2.7-code", "kimi-k2.6", "kimi-k2.5"],
+ "default_model": "kimi-k3",
+ "popular_models": ["kimi-k3"],
+ "_popular_note": "Only kimi-k3 meets the 1M-context policy (k2.x are 262144); k2.x remain reachable via live fetch_models.",
+ "_models_note": "kimi-k3 GA (platform.kimi.ai): 2.8T params, 1M context, vision + tool calling + structured outputs, $3/$15 per MTok ($0.30 cache-hit). Default max_completion_tokens 131072 (configurable up to the full window). k2.x kept as cheaper tiers.",
"detection_patterns": ["kimi", "moonshot"],
"base_url": "https://api.moonshot.ai/v1",
"models_endpoint": "https://api.moonshot.ai/v1/models",
@@ -254,6 +263,7 @@
"thinking_default_on": ["kimi-k2.6", "kimi-k2.5", "kimi-k2.7-code"],
"fixed_temperature": {"kimi-k2.6": 0.6, "kimi-k2.5": 0.6, "kimi-k2.7-code": 0.6},
"max_output_tokens": {
+ "kimi-k3": 131072,
"kimi-k2.6": 32768,
"kimi-k2.5": 98304,
"kimi-k2.7-code": 98304,
@@ -270,7 +280,8 @@
},
"mistral": {
"default_model": "mistral-large-latest",
- "popular_models": ["mistral-large-latest", "mistral-medium-latest", "mistral-small-latest", "codestral-latest"],
+ "popular_models": [],
+ "_popular_note": "Empty per the 1M-context policy — Mistral tops out at 262144 (large/medium). Models remain reachable via live fetch_models.",
"detection_patterns": ["mistral", "codestral"],
"base_url": "https://api.mistral.ai/v1",
"models_endpoint": "https://api.mistral.ai/v1/models",
diff --git a/server/config/model_registry.json b/server/config/model_registry.json
index 8ad689f1..30f56da7 100644
--- a/server/config/model_registry.json
+++ b/server/config/model_registry.json
@@ -1,14 +1,161 @@
{
- "_generated": "2026-07-19T18:45:02.760858+00:00",
+ "_generated": "2026-07-23T14:05:03.239022+00:00",
"_source": "openrouter",
- "_model_count": 338,
+ "_model_count": 342,
"models": {
+ "poolside/laguna-s-2.1": {
+ "id": "poolside/laguna-s-2.1",
+ "name": "Poolside: Laguna S 2.1",
+ "provider": "poolside",
+ "local_id": "laguna-s-2.1",
+ "context_length": 1048576,
+ "max_output_tokens": 131072,
+ "input_price_per_mtok": 0.1,
+ "output_price_per_mtok": 0.2,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "stop",
+ "temperature",
+ "tool_choice",
+ "tools"
+ ]
+ },
+ "poolside/laguna-s-2.1:free": {
+ "id": "poolside/laguna-s-2.1:free",
+ "name": "Poolside: Laguna S 2.1 (free)",
+ "provider": "poolside",
+ "local_id": "laguna-s-2.1:free",
+ "context_length": 262144,
+ "max_output_tokens": 32768,
+ "input_price_per_mtok": 0.0,
+ "output_price_per_mtok": 0.0,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "temperature",
+ "tool_choice",
+ "tools"
+ ]
+ },
+ "gemini/gemini-3.6-flash": {
+ "id": "google/gemini-3.6-flash",
+ "name": "Google: Gemini 3.6 Flash",
+ "provider": "gemini",
+ "local_id": "gemini-3.6-flash",
+ "context_length": 1048576,
+ "max_output_tokens": 65536,
+ "input_price_per_mtok": 1.5,
+ "output_price_per_mtok": 7.5,
+ "supports_thinking": true,
+ "thinking_type": "budget",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "reasoning_effort",
+ "response_format",
+ "seed",
+ "stop",
+ "structured_outputs",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_p"
+ ]
+ },
+ "gemini/gemini-3.5-flash-lite": {
+ "id": "google/gemini-3.5-flash-lite",
+ "name": "Google: Gemini 3.5 Flash Lite",
+ "provider": "gemini",
+ "local_id": "gemini-3.5-flash-lite",
+ "context_length": 1048576,
+ "max_output_tokens": 65536,
+ "input_price_per_mtok": 0.3,
+ "output_price_per_mtok": 2.5,
+ "supports_thinking": true,
+ "thinking_type": "budget",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "include_reasoning",
+ "max_tokens",
+ "reasoning",
+ "reasoning_effort",
+ "response_format",
+ "seed",
+ "stop",
+ "structured_outputs",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_p"
+ ]
+ },
+ "meituan/longcat-2.0": {
+ "id": "meituan/longcat-2.0",
+ "name": "Meituan: LongCat 2.0",
+ "provider": "meituan",
+ "local_id": "longcat-2.0",
+ "context_length": 1048756,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.3,
+ "output_price_per_mtok": 1.2,
+ "supports_thinking": false,
+ "thinking_type": "none",
+ "temperature_range": [
+ 0.0,
+ 2.0
+ ],
+ "is_reasoning_model": false,
+ "supported_parameters": [
+ "frequency_penalty",
+ "include_reasoning",
+ "logit_bias",
+ "max_tokens",
+ "min_p",
+ "presence_penalty",
+ "reasoning",
+ "repetition_penalty",
+ "seed",
+ "stop",
+ "temperature",
+ "tool_choice",
+ "tools",
+ "top_k",
+ "top_p"
+ ]
+ },
"thinkingmachines/inkling": {
"id": "thinkingmachines/inkling",
"name": "Thinking Machines: Inkling",
"provider": "thinkingmachines",
"local_id": "inkling",
- "context_length": 1048576,
+ "context_length": 524288,
"max_output_tokens": 0,
"input_price_per_mtok": 1.0,
"output_price_per_mtok": 4.05,
@@ -506,9 +653,9 @@
"provider": "tencent",
"local_id": "hy3",
"context_length": 262144,
- "max_output_tokens": 131072,
- "input_price_per_mtok": 0.2,
- "output_price_per_mtok": 0.8,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.14,
+ "output_price_per_mtok": 0.58,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -520,6 +667,7 @@
"frequency_penalty",
"include_reasoning",
"logit_bias",
+ "max_completion_tokens",
"max_tokens",
"min_p",
"presence_penalty",
@@ -537,40 +685,6 @@
"top_p"
]
},
- "tencent/hy3:free": {
- "id": "tencent/hy3:free",
- "name": "Tencent: Hy3 (free)",
- "provider": "tencent",
- "local_id": "hy3:free",
- "context_length": 262144,
- "max_output_tokens": 262144,
- "input_price_per_mtok": 0.0,
- "output_price_per_mtok": 0.0,
- "supports_thinking": false,
- "thinking_type": "none",
- "temperature_range": [
- 0.0,
- 2.0
- ],
- "is_reasoning_model": false,
- "supported_parameters": [
- "frequency_penalty",
- "include_reasoning",
- "max_tokens",
- "presence_penalty",
- "reasoning",
- "reasoning_effort",
- "repetition_penalty",
- "seed",
- "stop",
- "structured_outputs",
- "temperature",
- "tool_choice",
- "tools",
- "top_k",
- "top_p"
- ]
- },
"poolside/laguna-xs-2.1": {
"id": "poolside/laguna-xs-2.1",
"name": "Poolside: Laguna XS 2.1",
@@ -657,7 +771,7 @@
"provider": "gemini",
"local_id": "gemini-3.1-flash-lite-image",
"context_length": 65536,
- "max_output_tokens": 65536,
+ "max_output_tokens": 66000,
"input_price_per_mtok": 0.25,
"output_price_per_mtok": 1.5,
"supports_thinking": true,
@@ -768,7 +882,7 @@
"name": "Google: Nano Banana Pro (Gemini 3 Pro Image)",
"provider": "gemini",
"local_id": "gemini-3-pro-image",
- "context_length": 65536,
+ "context_length": 131072,
"max_output_tokens": 32768,
"input_price_per_mtok": 2.0,
"output_price_per_mtok": 12.0,
@@ -831,8 +945,8 @@
"local_id": "glm-5.2",
"context_length": 1048576,
"max_output_tokens": 131072,
- "input_price_per_mtok": 0.238,
- "output_price_per_mtok": 0.748,
+ "input_price_per_mtok": 0.7826,
+ "output_price_per_mtok": 2.4596,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -889,8 +1003,8 @@
"local_id": "kimi-k2.7-code",
"context_length": 262144,
"max_output_tokens": 262144,
- "input_price_per_mtok": 0.85,
- "output_price_per_mtok": 3.8,
+ "input_price_per_mtok": 0.82,
+ "output_price_per_mtok": 3.75,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -1041,10 +1155,10 @@
"name": "NVIDIA: Nemotron 3 Ultra",
"provider": "nvidia",
"local_id": "nemotron-3-ultra-550b-a55b",
- "context_length": 1000000,
- "max_output_tokens": 0,
- "input_price_per_mtok": 0.6,
- "output_price_per_mtok": 3.6,
+ "context_length": 512288,
+ "max_output_tokens": 16384,
+ "input_price_per_mtok": 0.5,
+ "output_price_per_mtok": 2.2,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -1176,7 +1290,7 @@
"name": "StepFun: Step 3.7 Flash",
"provider": "stepfun",
"local_id": "step-3.7-flash",
- "context_length": 256000,
+ "context_length": 262144,
"max_output_tokens": 256000,
"input_price_per_mtok": 0.2,
"output_price_per_mtok": 1.15,
@@ -1822,7 +1936,7 @@
"context_length": 1048576,
"max_output_tokens": 65536,
"input_price_per_mtok": 1.5,
- "output_price_per_mtok": 9.0,
+ "output_price_per_mtok": 7.5,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -2044,9 +2158,9 @@
"provider": "qwen",
"local_id": "qwen3.6-27b",
"context_length": 262144,
- "max_output_tokens": 65536,
- "input_price_per_mtok": 0.45,
- "output_price_per_mtok": 2.7,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.6,
+ "output_price_per_mtok": 3.6,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -2282,7 +2396,7 @@
"name": "Xiaomi: MiMo-V2.5-Pro",
"provider": "xiaomi",
"local_id": "mimo-v2.5-pro",
- "context_length": 1048576,
+ "context_length": 1050000,
"max_output_tokens": 131072,
"input_price_per_mtok": 0.435,
"output_price_per_mtok": 0.87,
@@ -2320,7 +2434,7 @@
"name": "Xiaomi: MiMo-V2.5",
"provider": "xiaomi",
"local_id": "mimo-v2.5",
- "context_length": 1048576,
+ "context_length": 1050000,
"max_output_tokens": 131072,
"input_price_per_mtok": 0.14,
"output_price_per_mtok": 0.28,
@@ -2542,7 +2656,7 @@
"name": "Z.ai: GLM 5.1",
"provider": "z-ai",
"local_id": "glm-5.1",
- "context_length": 202752,
+ "context_length": 204800,
"max_output_tokens": 128000,
"input_price_per_mtok": 0.966,
"output_price_per_mtok": 3.036,
@@ -2581,9 +2695,9 @@
"provider": "gemini",
"local_id": "gemma-4-26b-a4b-it",
"context_length": 262144,
- "max_output_tokens": 16384,
- "input_price_per_mtok": 0.07,
- "output_price_per_mtok": 0.34,
+ "max_output_tokens": 262144,
+ "input_price_per_mtok": 0.12,
+ "output_price_per_mtok": 0.35,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -2656,8 +2770,8 @@
"local_id": "gemma-4-31b-it",
"context_length": 262144,
"max_output_tokens": 262144,
- "input_price_per_mtok": 0.22,
- "output_price_per_mtok": 0.55,
+ "input_price_per_mtok": 0.12,
+ "output_price_per_mtok": 0.35,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -2793,13 +2907,23 @@
],
"is_reasoning_model": false,
"supported_parameters": [
+ "frequency_penalty",
"include_reasoning",
+ "logit_bias",
+ "logprobs",
"max_tokens",
+ "presence_penalty",
"reasoning",
+ "repetition_penalty",
+ "response_format",
+ "seed",
+ "stop",
+ "structured_outputs",
"temperature",
"tool_choice",
"tools",
"top_k",
+ "top_logprobs",
"top_p"
]
},
@@ -2917,7 +3041,7 @@
"name": "Kwaipilot: KAT-Coder-Pro V2",
"provider": "kwaipilot",
"local_id": "kat-coder-pro-v2",
- "context_length": 256000,
+ "context_length": 262144,
"max_output_tokens": 80000,
"input_price_per_mtok": 0.3,
"output_price_per_mtok": 1.2,
@@ -3144,9 +3268,9 @@
"provider": "nvidia",
"local_id": "nemotron-3-super-120b-a12b",
"context_length": 1000000,
- "max_output_tokens": 16384,
- "input_price_per_mtok": 0.085,
- "output_price_per_mtok": 0.4,
+ "max_output_tokens": 0,
+ "input_price_per_mtok": 0.08,
+ "output_price_per_mtok": 0.45,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -3182,7 +3306,7 @@
"name": "NVIDIA: Nemotron 3 Super (free)",
"provider": "nvidia",
"local_id": "nemotron-3-super-120b-a12b:free",
- "context_length": 1000000,
+ "context_length": 262144,
"max_output_tokens": 262144,
"input_price_per_mtok": 0.0,
"output_price_per_mtok": 0.0,
@@ -3456,7 +3580,7 @@
"provider": "gemini",
"local_id": "gemini-3.1-flash-image-preview",
"context_length": 131072,
- "max_output_tokens": 32768,
+ "max_output_tokens": 65536,
"input_price_per_mtok": 0.5,
"output_price_per_mtok": 3.0,
"supports_thinking": true,
@@ -3627,7 +3751,7 @@
"name": "Google: Gemini 3.1 Pro Preview Custom Tools",
"provider": "gemini",
"local_id": "gemini-3.1-pro-preview-customtools",
- "context_length": 1048756,
+ "context_length": 1048576,
"max_output_tokens": 65536,
"input_price_per_mtok": 2.0,
"output_price_per_mtok": 12.0,
@@ -3886,10 +4010,10 @@
"name": "Z.ai: GLM 5",
"provider": "z-ai",
"local_id": "glm-5",
- "context_length": 202752,
- "max_output_tokens": 202752,
+ "context_length": 204800,
+ "max_output_tokens": 131072,
"input_price_per_mtok": 0.95,
- "output_price_per_mtok": 3.15,
+ "output_price_per_mtok": 2.55,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -4266,9 +4390,9 @@
"name": "Z.ai: GLM 4.7 Flash",
"provider": "z-ai",
"local_id": "glm-4.7-flash",
- "context_length": 200000,
- "max_output_tokens": 131072,
- "input_price_per_mtok": 0.0605,
+ "context_length": 202752,
+ "max_output_tokens": 16384,
+ "input_price_per_mtok": 0.06,
"output_price_per_mtok": 0.4,
"supports_thinking": false,
"thinking_type": "none",
@@ -4426,7 +4550,7 @@
"name": "Z.ai: GLM 4.7",
"provider": "z-ai",
"local_id": "glm-4.7",
- "context_length": 202752,
+ "context_length": 204800,
"max_output_tokens": 131072,
"input_price_per_mtok": 0.4,
"output_price_per_mtok": 1.75,
@@ -4821,10 +4945,8 @@
"is_reasoning_model": false,
"supported_parameters": [
"frequency_penalty",
- "logprobs",
"max_tokens",
"presence_penalty",
- "repetition_penalty",
"response_format",
"seed",
"stop",
@@ -4832,7 +4954,6 @@
"temperature",
"tool_choice",
"tools",
- "top_logprobs",
"top_p"
]
},
@@ -4887,10 +5008,8 @@
"is_reasoning_model": false,
"supported_parameters": [
"frequency_penalty",
- "logprobs",
"max_tokens",
"presence_penalty",
- "repetition_penalty",
"response_format",
"seed",
"stop",
@@ -4898,7 +5017,6 @@
"temperature",
"tool_choice",
"tools",
- "top_logprobs",
"top_p"
]
},
@@ -5426,7 +5544,7 @@
"name": "Qwen: Qwen3 VL 32B Instruct",
"provider": "qwen",
"local_id": "qwen3-vl-32b-instruct",
- "context_length": 262144,
+ "context_length": 131072,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.104,
"output_price_per_mtok": 0.416,
@@ -5553,7 +5671,7 @@
"name": "Qwen: Qwen3 VL 8B Thinking",
"provider": "qwen",
"local_id": "qwen3-vl-8b-thinking",
- "context_length": 256000,
+ "context_length": 131072,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.117,
"output_price_per_mtok": 1.365,
@@ -5585,7 +5703,7 @@
"name": "Qwen: Qwen3 VL 8B Instruct",
"provider": "qwen",
"local_id": "qwen3-vl-8b-instruct",
- "context_length": 256000,
+ "context_length": 262144,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.117,
"output_price_per_mtok": 0.455,
@@ -5749,7 +5867,7 @@
"name": "Qwen: Qwen3 VL 30B A3B Thinking",
"provider": "qwen",
"local_id": "qwen3-vl-30b-a3b-thinking",
- "context_length": 131072,
+ "context_length": 262144,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.13,
"output_price_per_mtok": 1.56,
@@ -5784,9 +5902,9 @@
"provider": "qwen",
"local_id": "qwen3-vl-30b-a3b-instruct",
"context_length": 262144,
- "max_output_tokens": 32768,
- "input_price_per_mtok": 0.13,
- "output_price_per_mtok": 0.52,
+ "max_output_tokens": 16384,
+ "input_price_per_mtok": 0.15,
+ "output_price_per_mtok": 0.6,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -5847,7 +5965,7 @@
"name": "Z.ai: GLM 4.6",
"provider": "z-ai",
"local_id": "glm-4.6",
- "context_length": 202752,
+ "context_length": 204800,
"max_output_tokens": 131072,
"input_price_per_mtok": 0.5,
"output_price_per_mtok": 2.0,
@@ -6043,7 +6161,7 @@
"name": "Qwen: Qwen3 VL 235B A22B Instruct",
"provider": "qwen",
"local_id": "qwen3-vl-235b-a22b-instruct",
- "context_length": 131072,
+ "context_length": 262144,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.21,
"output_price_per_mtok": 1.9,
@@ -6166,7 +6284,7 @@
"name": "DeepSeek: DeepSeek V3.1 Terminus",
"provider": "deepseek",
"local_id": "deepseek-v3.1-terminus",
- "context_length": 131072,
+ "context_length": 163840,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.27,
"output_price_per_mtok": 1.0,
@@ -6269,8 +6387,8 @@
"local_id": "qwen3-next-80b-a3b-instruct",
"context_length": 262144,
"max_output_tokens": 262144,
- "input_price_per_mtok": 0.1,
- "output_price_per_mtok": 1.1,
+ "input_price_per_mtok": 0.15,
+ "output_price_per_mtok": 1.2,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -6424,7 +6542,7 @@
"name": "Qwen: Qwen3 30B A3B Thinking 2507",
"provider": "qwen",
"local_id": "qwen3-30b-a3b-thinking-2507",
- "context_length": 131072,
+ "context_length": 81920,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.13,
"output_price_per_mtok": 1.56,
@@ -6926,7 +7044,7 @@
"name": "Qwen: Qwen3 Coder 30B A3B Instruct",
"provider": "qwen",
"local_id": "qwen3-coder-30b-a3b-instruct",
- "context_length": 160000,
+ "context_length": 262144,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.07,
"output_price_per_mtok": 0.27,
@@ -7057,9 +7175,9 @@
"provider": "qwen",
"local_id": "qwen3-235b-a22b-thinking-2507",
"context_length": 262144,
- "max_output_tokens": 0,
- "input_price_per_mtok": 0.1495,
- "output_price_per_mtok": 1.495,
+ "max_output_tokens": 32768,
+ "input_price_per_mtok": 0.3,
+ "output_price_per_mtok": 3.0,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -7093,7 +7211,7 @@
"name": "Qwen: Qwen3 Coder 480B A35B",
"provider": "qwen",
"local_id": "qwen3-coder",
- "context_length": 1048576,
+ "context_length": 262144,
"max_output_tokens": 65536,
"input_price_per_mtok": 0.3,
"output_price_per_mtok": 1.0,
@@ -7360,7 +7478,7 @@
"name": "Baidu: ERNIE 4.5 VL 424B A47B ",
"provider": "baidu",
"local_id": "ernie-4.5-vl-424b-a47b",
- "context_length": 131072,
+ "context_length": 123000,
"max_output_tokens": 16000,
"input_price_per_mtok": 0.42,
"output_price_per_mtok": 1.25,
@@ -7390,7 +7508,7 @@
"name": "Mistral: Mistral Small 3.2 24B",
"provider": "mistralai",
"local_id": "mistral-small-3.2-24b-instruct",
- "context_length": 131072,
+ "context_length": 256000,
"max_output_tokens": 0,
"input_price_per_mtok": 0.1,
"output_price_per_mtok": 0.3,
@@ -7790,7 +7908,7 @@
"name": "Meta: Llama Guard 4 12B",
"provider": "meta",
"local_id": "llama-guard-4-12b",
- "context_length": 163840,
+ "context_length": 1048576,
"max_output_tokens": 16384,
"input_price_per_mtok": 0.18,
"output_price_per_mtok": 0.18,
@@ -7888,10 +8006,10 @@
"name": "Qwen: Qwen3 14B",
"provider": "qwen",
"local_id": "qwen3-14b",
- "context_length": 131702,
- "max_output_tokens": 16384,
- "input_price_per_mtok": 0.12,
- "output_price_per_mtok": 0.24,
+ "context_length": 131072,
+ "max_output_tokens": 8192,
+ "input_price_per_mtok": 0.2275,
+ "output_price_per_mtok": 0.91,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -8195,7 +8313,7 @@
"name": "Meta: Llama 4 Scout",
"provider": "meta",
"local_id": "llama-4-scout",
- "context_length": 10000000,
+ "context_length": 1310720,
"max_output_tokens": 16384,
"input_price_per_mtok": 0.1,
"output_price_per_mtok": 0.3,
@@ -8493,7 +8611,7 @@
"name": "Google: Gemma 3 27B",
"provider": "gemini",
"local_id": "gemma-3-27b-it",
- "context_length": 131072,
+ "context_length": 262144,
"max_output_tokens": 0,
"input_price_per_mtok": 0.1,
"output_price_per_mtok": 0.3,
@@ -8724,7 +8842,7 @@
"name": "Qwen: Qwen2.5 VL 72B Instruct",
"provider": "qwen",
"local_id": "qwen2.5-vl-72b-instruct",
- "context_length": 131072,
+ "context_length": 128000,
"max_output_tokens": 128000,
"input_price_per_mtok": 0.8,
"output_price_per_mtok": 1.0,
@@ -8872,7 +8990,7 @@
"name": "DeepSeek: R1 Distill Llama 70B",
"provider": "deepseek",
"local_id": "deepseek-r1-distill-llama-70b",
- "context_length": 128000,
+ "context_length": 8192,
"max_output_tokens": 8192,
"input_price_per_mtok": 0.8,
"output_price_per_mtok": 0.8,
@@ -8991,7 +9109,7 @@
"name": "DeepSeek: DeepSeek V3",
"provider": "deepseek",
"local_id": "deepseek-chat",
- "context_length": 131072,
+ "context_length": 163840,
"max_output_tokens": 16000,
"input_price_per_mtok": 0.2002,
"output_price_per_mtok": 0.8001,
@@ -9288,7 +9406,7 @@
"name": "Qwen2.5 Coder 32B Instruct",
"provider": "qwen",
"local_id": "qwen-2.5-coder-32b-instruct",
- "context_length": 128000,
+ "context_length": 32768,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.66,
"output_price_per_mtok": 1.0,
@@ -9351,7 +9469,7 @@
"name": "Magnum v4 72B",
"provider": "anthracite-org",
"local_id": "magnum-v4-72b",
- "context_length": 32768,
+ "context_length": 16384,
"max_output_tokens": 2048,
"input_price_per_mtok": 3.0,
"output_price_per_mtok": 5.0,
@@ -9386,7 +9504,7 @@
"name": "Qwen: Qwen2.5 7B Instruct",
"provider": "qwen",
"local_id": "qwen-2.5-7b-instruct",
- "context_length": 131072,
+ "context_length": 32768,
"max_output_tokens": 32768,
"input_price_per_mtok": 0.04,
"output_price_per_mtok": 0.1,
@@ -9501,7 +9619,7 @@
"name": "Meta: Llama 3.2 1B Instruct",
"provider": "meta",
"local_id": "llama-3.2-1b-instruct",
- "context_length": 131072,
+ "context_length": 60000,
"max_output_tokens": 60000,
"input_price_per_mtok": 0.027,
"output_price_per_mtok": 0.201,
@@ -9532,9 +9650,9 @@
"provider": "meta",
"local_id": "llama-3.2-3b-instruct",
"context_length": 131072,
- "max_output_tokens": 80000,
- "input_price_per_mtok": 0.0509,
- "output_price_per_mtok": 0.335,
+ "max_output_tokens": 131072,
+ "input_price_per_mtok": 0.05,
+ "output_price_per_mtok": 0.33,
"supports_thinking": false,
"thinking_type": "none",
"temperature_range": [
@@ -9564,7 +9682,7 @@
"name": "Qwen2.5 72B Instruct",
"provider": "qwen",
"local_id": "qwen-2.5-72b-instruct",
- "context_length": 131072,
+ "context_length": 32768,
"max_output_tokens": 16384,
"input_price_per_mtok": 0.36,
"output_price_per_mtok": 0.4,
@@ -10136,7 +10254,7 @@
"name": "WizardLM-2 8x22B",
"provider": "microsoft",
"local_id": "wizardlm-2-8x22b",
- "context_length": 65536,
+ "context_length": 65535,
"max_output_tokens": 8000,
"input_price_per_mtok": 0.62,
"output_price_per_mtok": 0.62,
@@ -10463,7 +10581,7 @@
"provider": "undi95",
"local_id": "remm-slerp-l2-13b",
"context_length": 6144,
- "max_output_tokens": 4096,
+ "max_output_tokens": 2048,
"input_price_per_mtok": 0.45,
"output_price_per_mtok": 0.65,
"supports_thinking": false,
@@ -10497,7 +10615,7 @@
"name": "MythoMax 13B",
"provider": "gryphe",
"local_id": "mythomax-l2-13b",
- "context_length": 4096,
+ "context_length": 8192,
"max_output_tokens": 4096,
"input_price_per_mtok": 0.06,
"output_price_per_mtok": 0.06,
diff --git a/server/config/pricing.json b/server/config/pricing.json
index c3479faf..b939a639 100644
--- a/server/config/pricing.json
+++ b/server/config/pricing.json
@@ -1,6 +1,6 @@
{
"version": "2026.07",
- "last_updated": "2026-07-11",
+ "last_updated": "2026-07-23",
"llm": {
"openai": {
@@ -38,6 +38,8 @@
"_default": {"input": 0.50, "output": 3.00}
},
"groq": {
+ "gpt-oss-120b": {"input": 0.15, "output": 0.60},
+ "gpt-oss-20b": {"input": 0.075, "output": 0.30},
"llama-4-scout": {"input": 0.11, "output": 0.34},
"llama-4-maverick": {"input": 0.15, "output": 0.60},
"llama-3.3-70b": {"input": 0.10, "output": 0.10},
@@ -45,6 +47,11 @@
"qwen3-32b": {"input": 0.29, "output": 0.59},
"_default": {"input": 0.15, "output": 0.60}
},
+ "xai": {
+ "grok-4.5": {"input": 2.00, "output": 6.00},
+ "grok-4.20": {"input": 2.00, "output": 6.00},
+ "_default": {"input": 2.00, "output": 6.00}
+ },
"cerebras": {
"gpt-oss-120b": {"input": 0.35, "output": 0.75},
"zai-glm-4.7": {"input": 2.25, "output": 2.75},
@@ -55,11 +62,14 @@
"_default": {"input": 1.00, "output": 5.00}
},
"deepseek": {
+ "deepseek-v4-pro": {"input": 0.435, "output": 0.87},
+ "deepseek-v4-flash": {"input": 0.14, "output": 0.28},
"deepseek-chat": {"input": 0.28, "output": 0.42, "cache_read": 0.028},
"deepseek-reasoner": {"input": 0.55, "output": 2.19, "cache_read": 0.14},
"_default": {"input": 0.28, "output": 0.42}
},
"kimi": {
+ "kimi-k3": {"input": 3.00, "output": 15.00, "cache_read": 0.30},
"kimi-k2.5": {"input": 0.60, "output": 2.50, "cache_read": 0.15},
"kimi-k2-thinking": {"input": 0.60, "output": 2.50, "cache_read": 0.15},
"moonshot-v1-8k": {"input": 0.20, "output": 2.00},
diff --git a/server/tests/llm/test_global_model_catalogue.py b/server/tests/llm/test_global_model_catalogue.py
index c8b02b7f..ceb61fbd 100644
--- a/server/tests/llm/test_global_model_catalogue.py
+++ b/server/tests/llm/test_global_model_catalogue.py
@@ -34,12 +34,31 @@ def test_openrouter_global_models_exist_in_latest_router_cache() -> None:
def test_latest_router_families_are_exposed() -> None:
+ # Global lists carry only >=1M-context models (llm_defaults _description
+ # policy, 2026-07-23) — hence grok-4.20 (2M) rather than the 500K grok-4.5.
providers = _defaults()["providers"]
assert "gpt-5.6-sol-pro" in providers["openai"]["popular_models"]
assert "claude-sonnet-5" in providers["anthropic"]["popular_models"]
assert "gemini-3.5-flash" in providers["gemini"]["popular_models"]
assert "kimi-k3" in providers["kimi"]["popular_models"]
- assert "grok-4.5" in providers["xai"]["popular_models"]
+ assert "grok-4.20" in providers["xai"]["popular_models"]
+
+
+def test_global_lists_only_carry_1m_context_models() -> None:
+ """Every popular_models entry must have a >=1M-token context window.
+
+ Context comes from the provider's own curated context_length map;
+ entries without a per-model value (e.g. openrouter's vendor-prefixed
+ ids) are exempt because their true windows live in the router cache.
+ """
+ providers = _defaults()["providers"]
+ for provider, config in providers.items():
+ context_map = config.get("context_length", {})
+ for model in config.get("popular_models", []):
+ ctx = context_map.get(model)
+ if ctx is None:
+ continue
+ assert ctx >= 1_000_000, f"{provider}/{model} has {ctx} < 1M context but is in the global list"
def test_ai_service_offline_fallback_uses_explicit_order() -> None: