@just-every/ensemble 0.2.268 → 0.2.270

package/dist/data/model_data.js

@@ -3,7 +3,7 @@ export const MODEL_CLASSES = {
     standard: {
         models: [
             'gpt-5.5',
-            'gemini-3-flash-preview',
+            'gemini-3.6-flash',
             'claude-sonnet-5',
             'grok-4.5',
         ],
@@ -12,7 +12,7 @@ export const MODEL_CLASSES = {
     mini: {
         models: [
             'gpt-5.4-nano',
-            'gemini-2.5-flash-lite',
+            'gemini-3.5-flash-lite',
             'claude-haiku-4-5-20251001',
             'grok-3-mini',
         ],
@@ -39,7 +39,7 @@ export const MODEL_CLASSES = {
     reasoning_mini: {
         models: [
             'gpt-5.4-mini',
-            'gemini-3-flash-preview',
+            'gemini-3.5-flash-lite',
             'claude-sonnet-5',
             'grok-3-mini',
         ],
@@ -76,7 +76,7 @@ export const MODEL_CLASSES = {
     writing: {
         models: [
             'gpt-5.5',
-            'gemini-3-flash-preview',
+            'gemini-3.6-flash',
             'claude-sonnet-5',
             'grok-4.5',
         ],
@@ -85,7 +85,7 @@ export const MODEL_CLASSES = {
     summary: {
         models: [
             'gpt-5-nano',
-            'gemini-2.5-flash-lite',
+            'gemini-3.5-flash-lite',
             'claude-haiku-4-5-20251001',
             'grok-3-mini',
         ],
@@ -103,7 +103,7 @@ export const MODEL_CLASSES = {
     vision_mini: {
         models: [
             'gpt-5.4-mini',
-            'gemini-3-flash-preview',
+            'gemini-3.5-flash-lite',
             'claude-haiku-4-5-20251001',
             'grok-3-mini',
         ],
@@ -135,13 +135,17 @@ export const MODEL_CLASSES = {
             'luma-photon-1',
             'ideogram-3.0',
             'midjourney-v7',
-            'grok-imagine-image-pro',
+            'grok-imagine-image-quality',
             'flux-kontext-pro',
             'stability-ultra',
             'runway-gen4-image',
             'recraft-v3',
         ],
     },
+    video_generation: {
+        models: ['grok-imagine-video', 'grok-imagine-video-1.5'],
+        description: 'Video generation and image-to-video models',
+    },
     embedding: {
         models: [
             'text-embedding-3-large',
@@ -152,7 +156,7 @@ export const MODEL_CLASSES = {
     voice: {
         models: [
             'tts-1-hd',
-            'eleven_multilingual_v2',
+            'eleven_v3',
             'gemini-3.1-flash-tts-preview',
         ],
         description: 'Text-to-Speech models for voice generation',
@@ -445,8 +449,8 @@ export const MODEL_REGISTRY = [
         id: 'meta-llama/llama-4-maverick',
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.15,
-            output_per_million: 0.6,
+            input_per_million: 0.2,
+            output_per_million: 0.8,
         },
         features: {
             context_length: 1048576,
@@ -470,11 +474,11 @@ export const MODEL_REGISTRY = [
         id: 'meta-llama/llama-4-scout',
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.08,
+            input_per_million: 0.1,
             output_per_million: 0.3,
         },
         features: {
-            context_length: 10000000,
+            context_length: 1310720,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
@@ -540,12 +544,27 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'grok-4.5',
-        aliases: ['grok-4.5-latest', 'grok-build-latest'],
+        aliases: ['grok-4.5-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 2.0,
-            output_per_million: 6.0,
-            cached_input_per_million: 0.5,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2,
+                price_above_threshold_per_million: 4,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.3,
+                price_above_threshold_per_million: 0.6,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 6,
+                price_above_threshold_per_million: 12,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
             context_length: 500_000,
@@ -561,10 +580,27 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'grok-4.3',
+        aliases: ['grok-4.3-latest', 'grok-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 1.25,
-            output_per_million: 2.5,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 1.25,
+                price_above_threshold_per_million: 2.5,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.2,
+                price_above_threshold_per_million: 0.4,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2.5,
+                price_above_threshold_per_million: 5,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
             context_length: 1_000_000,
@@ -579,11 +615,27 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'grok-build-0.1',
-        aliases: ['grok-build'],
+        aliases: ['grok-build', 'grok-build-latest', 'grok-code-fast-1', 'grok-code-fast'],
         provider: 'xai',
         cost: {
-            input_per_million: 1.0,
-            output_per_million: 2.0,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 1,
+                price_above_threshold_per_million: 2,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.2,
+                price_above_threshold_per_million: 0.4,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2,
+                price_above_threshold_per_million: 4,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
             context_length: 256000,
@@ -649,16 +701,31 @@ export const MODEL_REGISTRY = [
         description: 'Grok 4.1 Fast without reasoning. 2M context, flat pricing, text/image input.',
     },
     {
-        id: 'grok-4',
-        aliases: ['grok-4-2025-09-01', 'grok-4.20-0309-reasoning'],
+        id: 'grok-4.20-0309-reasoning',
+        aliases: ['grok-4', 'grok-4.20', 'grok-4.20-reasoning', 'grok-4.20-reasoning-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 2.0,
-            output_per_million: 6.0,
-            cached_input_per_million: 0.2,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 1.25,
+                price_above_threshold_per_million: 2.5,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.2,
+                price_above_threshold_per_million: 0.4,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2.5,
+                price_above_threshold_per_million: 5,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
-            context_length: 2_000_000,
+            context_length: 1_000_000,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
@@ -672,18 +739,34 @@ export const MODEL_REGISTRY = [
             code: 88,
             reasoning: 85,
         },
-        description: 'Grok-4 stable alias, currently resolving to the Grok 4.20 reasoning model with 2M context.',
+        description: 'Grok 4.20 reasoning model with 1M context and long-context pricing above 200k input tokens.',
     },
     {
         id: 'grok-4.20-0309-non-reasoning',
+        aliases: ['grok-4.20-non-reasoning', 'grok-4.20-non-reasoning-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 2.0,
-            output_per_million: 6.0,
-            cached_input_per_million: 0.2,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 1.25,
+                price_above_threshold_per_million: 2.5,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.2,
+                price_above_threshold_per_million: 0.4,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2.5,
+                price_above_threshold_per_million: 5,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
-            context_length: 2_000_000,
+            context_length: 1_000_000,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
@@ -701,14 +784,30 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'grok-4.20-multi-agent-0309',
+        aliases: ['grok-4.20-multi-agent', 'grok-4.20-multi-agent-latest'],
         provider: 'xai',
         cost: {
-            input_per_million: 2.0,
-            output_per_million: 6.0,
-            cached_input_per_million: 0.2,
+            input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 1.25,
+                price_above_threshold_per_million: 2.5,
+                tier_basis: 'input_tokens',
+            },
+            cached_input_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 0.2,
+                price_above_threshold_per_million: 0.4,
+                tier_basis: 'input_tokens',
+            },
+            output_per_million: {
+                threshold_tokens: 200_000,
+                price_below_threshold_per_million: 2.5,
+                price_above_threshold_per_million: 5,
+                tier_basis: 'input_tokens',
+            },
         },
         features: {
-            context_length: 2_000_000,
+            context_length: 1_000_000,
             input_modality: ['text', 'image'],
             output_modality: ['text'],
             tool_use: true,
@@ -2365,6 +2464,27 @@ export const MODEL_REGISTRY = [
         class: 'standard',
         description: 'Gemini 3 Flash Preview - fast multimodal model with 1M context window.',
     },
+    {
+        id: 'gemini-3.6-flash',
+        aliases: ['gemini-flash-latest', 'models/gemini-3.6-flash', 'models/gemini-flash-latest'],
+        provider: 'google',
+        cost: {
+            input_per_million: 1.5,
+            output_per_million: 7.5,
+            cached_input_per_million: 0.15,
+        },
+        features: {
+            context_length: 1048576,
+            input_modality: ['text', 'image', 'video', 'audio'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 65536,
+        },
+        class: 'standard',
+        description: 'Gemini 3.6 Flash - stable frontier multimodal model optimized for agentic execution, coding, spatial reasoning, and efficient tool use.',
+    },
     {
         id: 'gemini-3.5-flash',
         provider: 'google',
@@ -2386,6 +2506,27 @@ export const MODEL_REGISTRY = [
         score: 88,
         description: 'Gemini 3.5 Flash - fast multimodal reasoning model with 1M context and native thinking levels.',
     },
+    {
+        id: 'gemini-3.5-flash-lite',
+        aliases: ['gemini-flash-lite-latest', 'models/gemini-3.5-flash-lite', 'models/gemini-flash-lite-latest'],
+        provider: 'google',
+        cost: {
+            input_per_million: 0.3,
+            output_per_million: 2.5,
+            cached_input_per_million: 0.03,
+        },
+        features: {
+            context_length: 1048576,
+            input_modality: ['text', 'image', 'video', 'audio'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 65536,
+        },
+        class: 'reasoning_mini',
+        description: 'Gemini 3.5 Flash-Lite - stable low-latency multimodal model for high-throughput subagents, document parsing, extraction, and structured data tasks.',
+    },
     {
         id: 'gemini-3.1-flash-lite',
         aliases: ['gemini-3.1-flash-lite-preview'],
@@ -2729,6 +2870,63 @@ export const MODEL_REGISTRY = [
         class: 'voice',
         description: "OpenAI's advanced text-to-speech model with natural-sounding output. Supports customizable tone, style, and emotion through instructions. 85% cheaper than ElevenLabs with estimated $0.015/minute of audio.",
     },
+    {
+        id: 'gpt-realtime-2.1',
+        provider: 'openai',
+        cost: {
+            input_per_million: { text: 4, audio: 32, image: 5 },
+            cached_input_per_million: { text: 0.4 },
+            output_per_million: { text: 24, audio: 64 },
+        },
+        features: {
+            context_length: 128000,
+            max_output_tokens: 32000,
+            input_modality: ['text', 'audio', 'image'],
+            output_modality: ['text', 'audio'],
+            tool_use: true,
+            streaming: true,
+            reasoning_output: true,
+        },
+        class: 'voice',
+        description: 'OpenAI flagship GA Realtime model for low-latency text and audio conversations with tools.',
+    },
+    {
+        id: 'gpt-realtime-2.1-mini',
+        provider: 'openai',
+        cost: {
+            input_per_million: { text: 0.6, audio: 10, image: 0.8 },
+            cached_input_per_million: { text: 0.06 },
+            output_per_million: { text: 2.4, audio: 20 },
+        },
+        features: {
+            context_length: 128000,
+            max_output_tokens: 32000,
+            input_modality: ['text', 'audio', 'image'],
+            output_modality: ['text', 'audio'],
+            tool_use: true,
+            streaming: true,
+        },
+        class: 'voice',
+        description: 'OpenAI cost-efficient GA Realtime model for low-latency text and audio conversations.',
+    },
+    {
+        id: 'gpt-audio-1.5',
+        provider: 'openai',
+        cost: {
+            input_per_million: { text: 2.5, audio: 32 },
+            output_per_million: { text: 10, audio: 64 },
+        },
+        features: {
+            context_length: 128000,
+            max_output_tokens: 16384,
+            input_modality: ['text', 'audio'],
+            output_modality: ['text', 'audio'],
+            tool_use: true,
+            streaming: true,
+        },
+        class: 'voice',
+        description: 'OpenAI Chat Completions audio model supporting audio input/output, streaming, and tools.',
+    },
     {
         id: 'tts-1',
         provider: 'openai',
@@ -2770,27 +2968,30 @@ export const MODEL_REGISTRY = [
             input_modality: ['text'],
             output_modality: ['audio'],
             streaming: true,
+            max_input_characters: 10000,
         },
         class: 'voice',
         description: "ElevenLabs' multilingual high quality text-to-speech model supporting 29 languages with natural voice capabilities.",
     },
     {
-        id: 'eleven_turbo_v2_5',
+        id: 'eleven_v3',
         provider: 'elevenlabs',
         cost: {
-            input_per_million: 27.5,
+            input_per_million: 55,
             output_per_million: 0,
         },
         features: {
             input_modality: ['text'],
             output_modality: ['audio'],
             streaming: true,
+            max_input_characters: 5000,
         },
         class: 'voice',
-        description: "ElevenLabs' turbo model optimized for low-latency text-to-speech with high quality output.",
+        description: "ElevenLabs' most expressive multilingual text-to-speech model with audio tags and dialogue support.",
     },
     {
         id: 'eleven_flash_v2_5',
+        aliases: ['eleven_turbo_v2_5'],
         provider: 'elevenlabs',
         cost: {
             input_per_million: 27.5,
@@ -2800,6 +3001,7 @@ export const MODEL_REGISTRY = [
             input_modality: ['text'],
             output_modality: ['audio'],
             streaming: true,
+            max_input_characters: 40000,
         },
         class: 'voice',
         description: "ElevenLabs' fastest model optimized for ultra low-latency text-to-speech.",
@@ -3141,23 +3343,26 @@ export const MODEL_REGISTRY = [
         description: 'Midjourney v7 text-to-image (via KIE API; requires third-party API key).',
     },
     {
-        id: 'grok-imagine-image-pro',
+        id: 'grok-imagine-image-quality',
+        aliases: ['grok-imagine-image-pro'],
         provider: 'xai',
         cost: {
-            per_image: 0.07,
+            per_input_image: 0.01,
+            per_image_by_resolution: { '1k': 0.05, '2k': 0.07 },
         },
         features: {
             input_modality: ['text', 'image'],
             output_modality: ['image'],
         },
         class: 'image_generation',
-        description: 'xAI Grok Imagine Image Pro for premium text-to-image and image-guided image generation.',
+        description: 'xAI Grok Imagine Image Quality for premium 1K/2K image generation and editing.',
     },
     {
         id: 'grok-imagine-image',
         provider: 'xai',
         cost: {
             per_image: 0.02,
+            per_input_image: 0.002,
         },
         features: {
             input_modality: ['text', 'image'],
@@ -3166,6 +3371,36 @@ export const MODEL_REGISTRY = [
         class: 'image_generation',
         description: 'xAI Grok Imagine Image for lower-cost text-to-image and image-guided image generation.',
     },
+    {
+        id: 'grok-imagine-video',
+        provider: 'xai',
+        cost: {
+            per_input_image: 0.002,
+            per_input_second: 0.01,
+            per_second_by_resolution: { '480p': 0.05, '720p': 0.07 },
+        },
+        features: {
+            input_modality: ['text', 'image', 'video'],
+            output_modality: ['video'],
+        },
+        class: 'video_generation',
+        description: 'xAI Grok Imagine Video for text-to-video, image-to-video, and video editing up to 15 seconds.',
+    },
+    {
+        id: 'grok-imagine-video-1.5',
+        aliases: ['grok-imagine-video-1.5-preview'],
+        provider: 'xai',
+        cost: {
+            per_input_image: 0.01,
+            per_second_by_resolution: { '480p': 0.08, '720p': 0.14, '1080p': 0.25 },
+        },
+        features: {
+            input_modality: ['image'],
+            output_modality: ['video'],
+        },
+        class: 'video_generation',
+        description: 'xAI Grok Imagine Video 1.5 for premium image-to-video generation up to 1080p.',
+    },
     {
         id: 'imagen-2',
         provider: 'google',
@@ -3333,92 +3568,48 @@ export const MODEL_REGISTRY = [
         description: 'OpenAI Whisper transcription with full-turn output',
     },
     {
-        id: 'deepseek-chat',
-        aliases: ['deepseek-v3-0324'],
+        id: 'deepseek-v4-pro',
+        aliases: ['DeepSeek-V4', 'DeepSeek-V4-Pro', 'deepseek-v4'],
         provider: 'deepseek',
         cost: {
-            input_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 0.27,
-                off_peak_price_per_million: 0.135,
-            },
-            cached_input_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 0.07,
-                off_peak_price_per_million: 0.035,
-            },
-            output_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 1.1,
-                off_peak_price_per_million: 0.55,
-            },
+            input_per_million: 0.435,
+            cached_input_per_million: 0.003625,
+            output_per_million: 0.87,
         },
         features: {
-            context_length: 64000,
-            max_output_tokens: 8192,
+            context_length: 1_000_000,
+            max_output_tokens: 384_000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
             streaming: true,
             json_output: true,
+            reasoning_output: true,
         },
-        class: 'standard',
-        score: 75,
-        description: 'DeepSeek-V3 chat model with FIM completion support and time-based pricing',
+        class: 'reasoning',
+        description: 'DeepSeek V4 Pro direct API model with native tools, JSON output, and high/max reasoning.',
     },
     {
-        id: 'deepseek-reasoner',
-        aliases: ['deepseek-r1-0528'],
+        id: 'deepseek-v4-flash',
+        aliases: ['DeepSeek-V4-Flash'],
         provider: 'deepseek',
         cost: {
-            input_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 0.55,
-                off_peak_price_per_million: 0.1375,
-            },
-            cached_input_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 0.14,
-                off_peak_price_per_million: 0.035,
-            },
-            output_per_million: {
-                peak_utc_start_hour: 0,
-                peak_utc_start_minute: 30,
-                peak_utc_end_hour: 16,
-                peak_utc_end_minute: 30,
-                peak_price_per_million: 2.19,
-                off_peak_price_per_million: 0.5475,
-            },
+            input_per_million: 0.14,
+            cached_input_per_million: 0.0028,
+            output_per_million: 0.28,
         },
         features: {
-            context_length: 64000,
-            max_output_tokens: 64000,
+            context_length: 1_000_000,
+            max_output_tokens: 384_000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
-            simulate_tools: true,
             streaming: true,
             json_output: true,
             reasoning_output: true,
         },
         class: 'reasoning',
-        score: 85,
-        description: 'DeepSeek-R1 advanced reasoning model with extended output and time-based pricing',
+        description: 'DeepSeek V4 Flash direct API model with native tools and switchable thinking.',
     },
     {
         id: 'gpt-oss-120b',
@@ -3426,8 +3617,8 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'openai/gpt-oss-120b',
         cost: {
-            input_per_million: 0.03,
-            output_per_million: 0.15,
+            input_per_million: 0.037,
+            output_per_million: 0.17,
         },
         features: {
             context_length: 131072,
@@ -3448,8 +3639,8 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'openai/gpt-oss-20b',
         cost: {
-            input_per_million: 0.029,
-            output_per_million: 0.14,
+            input_per_million: 0.03,
+            output_per_million: 0.13,
         },
         features: {
             context_length: 131072,
@@ -3469,9 +3660,9 @@ export const MODEL_REGISTRY = [
         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,
+            input_per_million: 1.475,
+            cached_input_per_million: 0.295,
+            output_per_million: 4.425,
         },
         features: {
             context_length: 1000000,
@@ -3604,9 +3795,9 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3.6-27b',
         cost: {
-            input_per_million: 0.285,
+            input_per_million: 0.45,
             cached_input_per_million: 0.15,
-            output_per_million: 2.4,
+            output_per_million: 2.7,
         },
         features: {
             context_length: 262144,
@@ -3615,7 +3806,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 262140,
+            max_output_tokens: 65536,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3627,9 +3818,8 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3-235b-a22b-thinking-2507',
         cost: {
-            input_per_million: 0.1,
-            cached_input_per_million: 0.1,
-            output_per_million: 0.1,
+            input_per_million: 0.3,
+            output_per_million: 3,
         },
         features: {
             context_length: 262144,
@@ -3638,7 +3828,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 262144,
+            max_output_tokens: 32768,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3651,11 +3841,11 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'qwen/qwen3-coder',
         cost: {
-            input_per_million: 0.22,
-            output_per_million: 1.8,
+            input_per_million: 0.3,
+            output_per_million: 1,
         },
         features: {
-            context_length: 1048576,
+            context_length: 262144,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -3678,7 +3868,7 @@ export const MODEL_REGISTRY = [
             cached_input_per_million: 0.1794,
         },
         features: {
-            context_length: 202752,
+            context_length: 204800,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -3695,9 +3885,9 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'z-ai/glm-5.2',
         cost: {
-            input_per_million: 0.93,
-            output_per_million: 3.0,
-            cached_input_per_million: 0.18,
+            input_per_million: 0.7756,
+            output_per_million: 2.4376,
+            cached_input_per_million: 0.14404,
         },
         features: {
             context_length: 1048576,
@@ -3706,7 +3896,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 32768,
+            max_output_tokens: 131072,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3714,7 +3904,6 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'deepseek/deepseek-v4-pro',
-        aliases: ['DeepSeek-V4', 'DeepSeek-V4-Pro', 'deepseek-v4', 'deepseek-v4-pro'],
         provider: 'openrouter',
         openrouter_id: 'deepseek/deepseek-v4-pro',
         cost: {
@@ -3737,13 +3926,12 @@ export const MODEL_REGISTRY = [
     },
     {
         id: 'deepseek/deepseek-v4-flash',
-        aliases: ['DeepSeek-V4-Flash', 'deepseek-v4-flash'],
         provider: 'openrouter',
         openrouter_id: 'deepseek/deepseek-v4-flash',
         cost: {
-            input_per_million: 0.09,
-            cached_input_per_million: 0.018,
-            output_per_million: 0.18,
+            input_per_million: 0.0938,
+            cached_input_per_million: 0.01876,
+            output_per_million: 0.1876,
         },
         features: {
             context_length: 1048576,
@@ -3752,7 +3940,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 131072,
+            max_output_tokens: 384000,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3764,12 +3952,12 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'xiaomi/mimo-v2.5',
         cost: {
-            input_per_million: 0.105,
-            cached_input_per_million: 0.028,
+            input_per_million: 0.14,
+            cached_input_per_million: 0.0028,
             output_per_million: 0.28,
         },
         features: {
-            context_length: 1048576,
+            context_length: 1050000,
             input_modality: ['text', 'audio', 'image', 'video'],
             output_modality: ['text'],
             tool_use: true,
@@ -3792,7 +3980,7 @@ export const MODEL_REGISTRY = [
             output_per_million: 0.87,
         },
         features: {
-            context_length: 1048576,
+            context_length: 1050000,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -3834,9 +4022,9 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'tencent/hy3',
         cost: {
-            input_per_million: 0.2,
-            cached_input_per_million: 0.5,
-            output_per_million: 0.8,
+            input_per_million: 0.14,
+            cached_input_per_million: 0.035,
+            output_per_million: 0.58,
         },
         features: {
             context_length: 262144,
@@ -3845,7 +4033,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: false,
-            max_output_tokens: 131072,
+            max_output_tokens: 262144,
             reasoning_output: true,
         },
         class: 'reasoning',
@@ -3856,9 +4044,9 @@ export const MODEL_REGISTRY = [
         aliases: ['kimi-k2.5', 'kimi-k2-5'],
         provider: 'openrouter',
         cost: {
-            input_per_million: 0.375,
-            output_per_million: 2.025,
-            cached_input_per_million: 0.203,
+            input_per_million: 0.57,
+            output_per_million: 2.85,
+            cached_input_per_million: 0.095,
         },
         features: {
             context_length: 262144,
@@ -3879,9 +4067,9 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'moonshotai/kimi-k2.6',
         cost: {
-            input_per_million: 0.66,
-            output_per_million: 3.41,
-            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: 262144,
@@ -3902,9 +4090,9 @@ export const MODEL_REGISTRY = [
         provider: 'openrouter',
         openrouter_id: 'moonshotai/kimi-k2.7-code',
         cost: {
-            input_per_million: 0.74,
-            output_per_million: 3.5,
-            cached_input_per_million: 0.15,
+            input_per_million: 0.82,
+            output_per_million: 3.75,
+            cached_input_per_million: 0.16,
         },
         features: {
             context_length: 262144,
@@ -3913,7 +4101,7 @@ export const MODEL_REGISTRY = [
             tool_use: true,
             streaming: true,
             json_output: true,
-            max_output_tokens: 16384,
+            max_output_tokens: 262144,
             reasoning_output: true,
         },
         class: 'code',
@@ -3953,7 +4141,7 @@ export const MODEL_REGISTRY = [
             output_per_million: 1.2,
         },
         features: {
-            context_length: 1_048_576,
+            context_length: 1_048_756,
             input_modality: ['text'],
             output_modality: ['text'],
             tool_use: true,
@@ -3976,7 +4164,7 @@ export const MODEL_REGISTRY = [
             output_per_million: 4.05,
         },
         features: {
-            context_length: 1_048_576,
+            context_length: 524_288,
             input_modality: ['text', 'image', 'audio'],
             output_modality: ['text'],
             tool_use: true,
@@ -4056,6 +4244,28 @@ export const MODEL_REGISTRY = [
         class: 'code',
         description: 'Kwaipilot KAT-Coder Air V2.5 via OpenRouter. Cost-efficient agentic coding model for autonomous repository changes and tool-driven workflows.',
     },
+    {
+        id: 'poolside/laguna-s-2.1',
+        aliases: ['Laguna S 2.1', 'laguna-s-2.1', 'laguna-2.1'],
+        provider: 'openrouter',
+        openrouter_id: 'poolside/laguna-s-2.1',
+        cost: {
+            input_per_million: 0.1,
+            cached_input_per_million: 0.01,
+            output_per_million: 0.2,
+        },
+        features: {
+            context_length: 1_048_576,
+            input_modality: ['text'],
+            output_modality: ['text'],
+            tool_use: true,
+            streaming: true,
+            json_output: true,
+            max_output_tokens: 131_072,
+        },
+        class: 'code',
+        description: 'Poolside Laguna S 2.1 via OpenRouter, optimized for long-context agentic coding workflows.',
+    },
 ];
 export function findModel(modelId) {
     const externalModel = getExternalModel(modelId);