@just-every/ensemble 0.2.245 → 0.2.247

package/dist/cjs/data/model_data.cjs

@@ -35,7 +35,7 @@ exports.MODEL_CLASSES = {
         models: [
             'gpt-5.5-pro',
             'gemini-3.1-pro-preview',
-            'claude-opus-4-7',
+            'claude-opus-4-8',
             'grok-4.3',
         ],
         random: true,
@@ -62,7 +62,7 @@ exports.MODEL_CLASSES = {
         models: [
             'gpt-5.5',
             'gemini-3.1-pro-preview',
-            'claude-opus-4-7',
+            'claude-opus-4-8',
             'grok-4.3',
         ],
         random: true,
@@ -71,7 +71,7 @@ exports.MODEL_CLASSES = {
         models: [
             'gpt-5.3-codex',
             'gemini-3.1-pro-preview',
-            'claude-opus-4-7',
+            'claude-opus-4-8',
             'grok-4.3',
             'qwen3-coder',
         ],
@@ -99,7 +99,7 @@ exports.MODEL_CLASSES = {
         models: [
             'gpt-5.5',
             'gemini-3.1-pro-preview',
-            'claude-opus-4-7',
+            'claude-opus-4-8',
             'grok-4.3',
         ],
         random: true,
@@ -124,7 +124,7 @@ exports.MODEL_CLASSES = {
         models: [
             'gpt-5.5',
             'gemini-3.1-pro-preview',
-            'claude-opus-4-7',
+            'claude-opus-4-8',
             'grok-4.3',
         ],
         random: true,
@@ -403,7 +403,7 @@ exports.MODEL_REGISTRY = [
         id: 'meta-llama/llama-4-maverick',
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.18,
+            input_per_million: 0.15,
             output_per_million: 0.6,
         },
         features: {
@@ -413,6 +413,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
+            max_output_tokens: 16384,
         },
         class: 'standard',
         score: 78,
@@ -431,12 +432,13 @@ exports.MODEL_REGISTRY = [
             output_per_million: 0.3,
         },
         features: {
-            context_length: 327680,
-            input_modality: ['text'],
+            context_length: 10000000,
+            input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
+            max_output_tokens: 16384,
         },
         class: 'mini',
         score: 65,
@@ -446,16 +448,17 @@ exports.MODEL_REGISTRY = [
         id: 'qwen/qwen3-235b-a22b',
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.1,
-            output_per_million: 0.1,
+            input_per_million: 0.455,
+            output_per_million: 1.82,
         },
         features: {
-            context_length: 40960,
+            context_length: 131072,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
+            max_output_tokens: 8192,
         },
         class: 'reasoning',
         score: 83,
@@ -466,30 +469,6 @@ exports.MODEL_REGISTRY = [
         },
         description: 'Qwen3-235B-A22B is a 235B parameter mixture-of-experts (MoE) model developed by Qwen, activating 22B parameters per forward pass.',
     },
-    {
-        id: 'qwen/qwen-max',
-        provider: 'openrouter',
-        cost: {
-            input_per_million: 1.6,
-            output_per_million: 6.4,
-        },
-        features: {
-            context_length: 131072,
-            input_modality: ['text', 'image'],
-            output_modality: ['text'],
-            tool_use: true,
-            streaming: true,
-            json_output: true,
-        },
-        class: 'reasoning',
-        score: 80,
-        scores: {
-            monologue: 73,
-            code: 61,
-            reasoning: 57,
-        },
-        description: 'Qwen-Max, based on Qwen2.5, provides the best inference performance among Qwen models, especially for complex multi-step tasks.',
-    },
     {
         id: 'qwen/qwen3.5-397b-a17b',
         aliases: ['qwen3.5-397b-a17b'],
@@ -505,7 +484,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 81920,
+            max_output_tokens: 65536,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -518,41 +497,41 @@ exports.MODEL_REGISTRY = [
         description: 'Qwen3.5-397B-A17B is a vision-language 397B MoE model with a native 262k-context window, multimodal input support, and strong reasoning.',
     },
     {
-        id: 'mistral/ministral-8b',
-        provider: 'openrouter',
+        id: 'grok-4.3',
+        provider: 'xai',
         cost: {
-            input_per_million: 0.1,
-            output_per_million: 0.1,
+            input_per_million: 1.25,
+            output_per_million: 2.5,
         },
         features: {
-            context_length: 131072,
-            input_modality: ['text'],
+            context_length: 1_000_000,
+            input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
         },
-        class: 'standard',
-        score: 55,
-        description: 'Ministral 8B is a state-of-the-art language model optimized for on-device and edge computing. Designed for efficiency in knowledge-intensive tasks, commonsense reasoning, and function-calling.',
+        class: 'reasoning',
+        description: "Grok 4.3, xAI's latest recommended reasoning model with 1M context, multimodal input, streaming, tools, structured output, and low/medium/high reasoning effort.",
     },
     {
-        id: 'grok-4.3',
+        id: 'grok-build-0.1',
+        aliases: ['grok-build', 'grok-build-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 1.25,
-            output_per_million: 2.5,
+            input_per_million: 1.0,
+            output_per_million: 2.0,
         },
         features: {
-            context_length: 1_000_000,
+            context_length: 256000,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
         },
-        class: 'reasoning',
-        description: "Grok 4.3, xAI's latest recommended reasoning model with 1M context, multimodal input, streaming, tools, structured output, and low/medium/high reasoning effort.",
+        class: 'code',
+        description: "Grok Build 0.1, xAI's fast coding model trained for agentic coding workflows with 256k context and text/image input.",
     },
     {
         id: 'grok-4-1-fast-reasoning',
@@ -1374,6 +1353,48 @@ exports.MODEL_REGISTRY = [
         score: 91,
         description: 'ChatGPT-optimized GPT-5.1 chat model (128k/16k).',
     },
+    {
+        id: 'codex-gpt-5.5',
+        aliases: ['codex-gpt-5.5-2026-04-23'],
+        provider: 'codex',
+        cost: {
+            input_per_million: {
+                threshold_tokens: 272000,
+                price_below_threshold_per_million: 5.0,
+                price_above_threshold_per_million: 10.0,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 272000,
+                price_below_threshold_per_million: 0.5,
+                price_above_threshold_per_million: 1.0,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 272000,
+                price_below_threshold_per_million: 30.0,
+                price_above_threshold_per_million: 45.0,
+                tier_basis: 'input_tokens',
+            },
+        },
+        features: {
+            context_length: 1050000,
+            max_output_tokens: 128000,
+            input_modality: ['text', 'image'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: false,
+            json_output: true,
+        },
+        class: 'code',
+        score: 99,
+        scores: {
+            monologue: 99,
+            code: 99,
+            reasoning: 98,
+        },
+        description: 'Codex CLI GPT-5.5 text model for agentic coding, visual repair, and tool-mediated workflows.',
+    },
     {
         id: 'gpt-5-codex',
         aliases: ['gpt-5-codex-2025-09-15'],
@@ -1779,7 +1800,7 @@ exports.MODEL_REGISTRY = [
     },
     {
         id: 'claude-sonnet-4-6',
-        aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'claude-sonnet-latest'],
+        aliases: ['claude-sonnet-4-6', 'claude-sonnet-4.6', 'claude-sonnet-latest', 'sonnet-4-6', 'sonnet-4.6'],
         provider: 'anthropic',
         cost: {
             input_per_million: 3.0,
@@ -1978,7 +1999,7 @@ exports.MODEL_REGISTRY = [
     },
     {
         id: 'claude-opus-4-7',
-        aliases: ['claude-opus-4-7', 'claude-opus-4.7', 'claude-opus', 'claude-opus-latest', 'claude-4.7-opus'],
+        aliases: ['claude-opus-4-7', 'claude-opus-4.7', 'claude-4.7-opus'],
         provider: 'anthropic',
         cost: {
             input_per_million: 5.0,
@@ -1998,6 +2019,36 @@ exports.MODEL_REGISTRY = [
         class: 'reasoning',
         description: 'Claude Opus 4.7 - Frontier Opus model for complex reasoning, agentic coding, and high-resolution vision.',
     },
+    {
+        id: 'claude-opus-4-8',
+        aliases: [
+            'claude-opus-4-8',
+            'claude-opus-4.8',
+            'claude-opus',
+            'claude-opus-latest',
+            'claude-4.8-opus',
+            'opus-4-8',
+            'opus-4.8',
+        ],
+        provider: 'anthropic',
+        cost: {
+            input_per_million: 5.0,
+            output_per_million: 25.0,
+            cached_input_per_million: 0.5,
+        },
+        features: {
+            context_length: 1_000_000,
+            input_modality: ['text', 'image'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            reasoning_output: true,
+            max_output_tokens: 128000,
+        },
+        class: 'reasoning',
+        description: 'Claude Opus 4.8 - Frontier Opus model for complex reasoning, agentic coding, and high-resolution vision.',
+    },
     {
         id: 'gemini-2.5-pro-preview-06-05',
         aliases: ['gemini-2.5-pro', 'gemini-2.5-pro-exp-03-25', 'gemini-2.5-pro-preview-05-06'],
@@ -2067,7 +2118,8 @@ exports.MODEL_REGISTRY = [
         description: 'Gemini 3.5 Flash - fast multimodal reasoning model with 1M context and native thinking levels.',
     },
     {
-        id: 'gemini-3.1-flash-lite-preview',
+        id: 'gemini-3.1-flash-lite',
+        aliases: ['gemini-3.1-flash-lite-preview'],
         provider: 'google',
         cost: {
             input_per_million: {
@@ -2076,6 +2128,7 @@ exports.MODEL_REGISTRY = [
                 video: 0.25,
                 audio: 0.5,
             },
+            cached_input_per_million: 0.025,
             output_per_million: 1.5,
         },
         features: {
@@ -2089,7 +2142,7 @@ exports.MODEL_REGISTRY = [
         },
         class: 'reasoning_mini',
         score: 82,
-        description: 'Gemini 3.1 Flash Lite Preview - cost-efficient multimodal model with 1M context window.',
+        description: 'Gemini 3.1 Flash Lite - stable cost-efficient multimodal model with 1M context window; preview ID remains supported as an alias.',
     },
     {
         id: 'gemini-2.5-flash',
@@ -2537,7 +2590,7 @@ exports.MODEL_REGISTRY = [
             output_per_million: 1.0,
         },
         features: {
-            context_length: 32768,
+            context_length: 127072,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -2555,34 +2608,17 @@ exports.MODEL_REGISTRY = [
             output_per_million: 15.0,
         },
         features: {
-            context_length: 32768,
+            context_length: 200000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
+            max_output_tokens: 8000,
         },
         class: 'standard',
         description: 'Advanced search model optimized for complex queries and deeper content understanding.',
     },
-    {
-        id: 'perplexity/sonar-reasoning',
-        provider: 'openrouter',
-        cost: {
-            input_per_million: 1.0,
-            output_per_million: 5.0,
-        },
-        features: {
-            context_length: 32768,
-            input_modality: ['text'],
-            output_modality: ['text'],
-            tool_use: true,
-            streaming: true,
-            json_output: true,
-        },
-        class: 'reasoning_mini',
-        description: 'Quick problem-solving and reasoning model, ideal for evaluating complex queries.',
-    },
     {
         id: 'perplexity/sonar-reasoning-pro',
         provider: 'openrouter',
@@ -2591,7 +2627,7 @@ exports.MODEL_REGISTRY = [
             output_per_million: 8.0,
         },
         features: {
-            context_length: 32768,
+            context_length: 128000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -2609,7 +2645,7 @@ exports.MODEL_REGISTRY = [
             output_per_million: 8.0,
         },
         features: {
-            context_length: 32768,
+            context_length: 128000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -2619,45 +2655,6 @@ exports.MODEL_REGISTRY = [
         class: 'standard',
         description: 'Best suited for exhaustive research, generating detailed reports and in-depth insights.',
     },
-    {
-        id: 'mistralai/magistral-small-2506',
-        provider: 'openrouter',
-        cost: {
-            input_per_million: 0.5,
-            output_per_million: 1.5,
-        },
-        features: {
-            context_length: 40000,
-            input_modality: ['text'],
-            output_modality: ['text'],
-            tool_use: true,
-            streaming: true,
-            json_output: true,
-        },
-        class: 'reasoning_mini',
-        score: 72,
-        description: 'Magistral Small is a 24B parameter instruction-tuned model based on Mistral-Small-3.1 (2503), enhanced through supervised fine-tuning on traces from Magistral Medium and further refined via reinforcement learning. It is optimized for reasoning and supports a wide multilingual range, including over 20 languages.',
-    },
-    {
-        id: 'mistralai/magistral-medium-2506:thinking',
-        provider: 'openrouter',
-        cost: {
-            input_per_million: 2.0,
-            output_per_million: 5.0,
-        },
-        features: {
-            context_length: 40960,
-            input_modality: ['text'],
-            output_modality: ['text'],
-            tool_use: true,
-            streaming: true,
-            json_output: true,
-            reasoning_output: true,
-        },
-        class: 'reasoning',
-        score: 80,
-        description: "Magistral is Mistral's first reasoning model. It is ideal for general purpose use requiring longer thought processing and better accuracy than with non-reasoning LLMs. From legal research and financial forecasting to software development and creative storytelling — this model solves multi-step challenges where transparency and precision are critical.",
-    },
     {
         id: 'test-model',
         provider: 'test',
@@ -3053,8 +3050,8 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'openai/gpt-oss-120b',
         cost: {
-            input_per_million: 0.1,
-            output_per_million: 0.5,
+            input_per_million: 0.039,
+            output_per_million: 0.18,
         },
         features: {
             context_length: 131072,
@@ -3063,7 +3060,6 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 100000,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3076,8 +3072,8 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'openai/gpt-oss-20b',
         cost: {
-            input_per_million: 0.05,
-            output_per_million: 0.2,
+            input_per_million: 0.029,
+            output_per_million: 0.14,
         },
         features: {
             context_length: 131072,
@@ -3086,12 +3082,34 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 100000,
         },
         class: 'standard',
         score: 82,
         description: 'GPT OSS 20B - MoE model with 3.6B active params, optimized for consumer hardware',
     },
+    {
+        id: 'qwen/qwen3.7-max',
+        aliases: ['Qwen3.7 Max', 'qwen3.7-max', 'qwen-3.7-max'],
+        provider: 'openrouter',
+        openrouter_id: 'qwen/qwen3.7-max',
+        cost: {
+            input_per_million: 1.25,
+            cached_input_per_million: 0.25,
+            output_per_million: 3.75,
+        },
+        features: {
+            context_length: 1000000,
+            input_modality: ['text'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 65536,
+            reasoning_output: true,
+        },
+        class: 'code',
+        description: 'Qwen3.7 Max via OpenRouter. Flagship Qwen 3.7 model for agent-centric coding, productivity, and long-horizon autonomous execution.',
+    },
     {
         id: 'qwen/qwen3.6-plus',
         aliases: ['Qwen 3.6', 'Qwen3.6', 'qwen3.6', 'qwen-3.6', 'qwen3.6-plus', 'qwen-3.6-plus'],
@@ -3120,9 +3138,8 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3.6-flash',
         cost: {
-            input_per_million: 0.25,
-            cached_input_per_million: 0.3125,
-            output_per_million: 1.5,
+            input_per_million: 0.1875,
+            output_per_million: 1.125,
         },
         features: {
             context_length: 1000000,
@@ -3143,9 +3160,8 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3.6-35b-a3b',
         cost: {
-            input_per_million: 0.1612,
-            cached_input_per_million: 0.1612,
-            output_per_million: 0.96525,
+            input_per_million: 0.14,
+            output_per_million: 1.0,
         },
         features: {
             context_length: 262144,
@@ -3154,7 +3170,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 65536,
+            max_output_tokens: 262140,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3189,7 +3205,7 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3.6-27b',
         cost: {
-            input_per_million: 0.32,
+            input_per_million: 0.29,
             output_per_million: 3.2,
         },
         features: {
@@ -3199,7 +3215,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 81920,
+            max_output_tokens: 262140,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3211,8 +3227,9 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3-235b-a22b-thinking-2507',
         cost: {
-            input_per_million: 0.078,
-            output_per_million: 0.312,
+            input_per_million: 0.1,
+            cached_input_per_million: 0.1,
+            output_per_million: 0.1,
         },
         features: {
             context_length: 262144,
@@ -3221,7 +3238,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 81920,
+            max_output_tokens: 262144,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3234,17 +3251,17 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3-coder',
         cost: {
-            input_per_million: 0.2,
-            output_per_million: 0.8,
+            input_per_million: 0.22,
+            output_per_million: 1.8,
         },
         features: {
-            context_length: 262144,
+            context_length: 1048576,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 100000,
+            max_output_tokens: 65536,
         },
         class: 'code',
         score: 90,
@@ -3256,9 +3273,9 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'z-ai/glm-5.1',
         cost: {
-            input_per_million: 1.05,
-            output_per_million: 3.5,
-            cached_input_per_million: 0.525,
+            input_per_million: 0.98,
+            output_per_million: 3.08,
+            cached_input_per_million: 0.182,
         },
         features: {
             context_length: 202752,
@@ -3267,7 +3284,6 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 65535,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3301,11 +3317,57 @@ exports.MODEL_REGISTRY = [
         aliases: ['DeepSeek-V4-Flash', 'deepseek-v4-flash'],
         provider: 'openrouter',
         openrouter_id: 'deepseek/deepseek-v4-flash',
+        cost: {
+            input_per_million: 0.0983,
+            cached_input_per_million: 0.0197,
+            output_per_million: 0.1966,
+        },
+        features: {
+            context_length: 1048576,
+            input_modality: ['text'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 131072,
+            reasoning_output: true,
+        },
+        class: 'reasoning',
+        description: 'DeepSeek V4 Flash via OpenRouter. Efficiency-optimized 284B-parameter MoE model with 13B active parameters and 1M context.',
+    },
+    {
+        id: 'xiaomi/mimo-v2.5',
+        aliases: ['MiMo-V2.5', 'mimo-v2.5', 'mimo-v2-5'],
+        provider: 'openrouter',
+        openrouter_id: 'xiaomi/mimo-v2.5',
         cost: {
             input_per_million: 0.14,
             cached_input_per_million: 0.0028,
             output_per_million: 0.28,
         },
+        features: {
+            context_length: 1048576,
+            input_modality: ['text', 'audio', 'image', 'video'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 131072,
+            reasoning_output: true,
+        },
+        class: 'reasoning',
+        description: 'Xiaomi MiMo V2.5 via OpenRouter. Native omnimodal 1M-context model for agentic workflows, multimodal perception, and cost-efficient reasoning.',
+    },
+    {
+        id: 'xiaomi/mimo-v2.5-pro',
+        aliases: ['MiMo-V2.5-Pro', 'mimo-v2.5-pro', 'mimo-v2-5-pro'],
+        provider: 'openrouter',
+        openrouter_id: 'xiaomi/mimo-v2.5-pro',
+        cost: {
+            input_per_million: 0.435,
+            cached_input_per_million: 0.0036,
+            output_per_million: 0.87,
+        },
         features: {
             context_length: 1048576,
             input_modality: ['text'],
@@ -3313,11 +3375,11 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 384000,
+            max_output_tokens: 131072,
             reasoning_output: true,
         },
         class: 'reasoning',
-        description: 'DeepSeek V4 Flash via OpenRouter. Efficiency-optimized 284B-parameter MoE model with 13B active parameters and 1M context.',
+        description: 'Xiaomi MiMo V2.5 Pro via OpenRouter. Flagship 1M-context model for agentic coding, complex software engineering, and long-horizon tasks.',
     },
     {
         id: 'tencent/hy3-preview',
@@ -3325,9 +3387,9 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'tencent/hy3-preview',
         cost: {
-            input_per_million: 0.066,
-            cached_input_per_million: 0.029,
-            output_per_million: 0.26,
+            input_per_million: 0.063,
+            cached_input_per_million: 0.021,
+            output_per_million: 0.21,
         },
         features: {
             context_length: 262144,
@@ -3336,7 +3398,6 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: false,
-            max_output_tokens: 262144,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3347,9 +3408,9 @@ exports.MODEL_REGISTRY = [
         aliases: ['kimi-k2.5', 'kimi-k2-5'],
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.45,
-            output_per_million: 2.2,
-            cached_input_per_million: 0.225,
+            input_per_million: 0.4,
+            output_per_million: 1.9,
+            cached_input_per_million: 0.09,
         },
         features: {
             context_length: 262144,
@@ -3358,6 +3419,7 @@ exports.MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
+            max_output_tokens: 262144,
         },
         class: 'reasoning',
         score: 86,
@@ -3369,18 +3431,18 @@ exports.MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'moonshotai/kimi-k2.6',
         cost: {
-            input_per_million: 0.74,
-            output_per_million: 3.49,
-            cached_input_per_million: 0.14,
+            input_per_million: 0.684,
+            output_per_million: 3.42,
+            cached_input_per_million: 0.144,
         },
         features: {
-            context_length: 262142,
+            context_length: 262144,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 262142,
+            max_output_tokens: 262144,
             reasoning_output: true,
         },
         class: 'reasoning',