🐛 fix: Add ExtendParamsTypeSchema for enhanced model settings (#11437)

* 🐛 fix: Update reasoning handling in OpenRouter and VercelAIGateway to include thinkingLevel and adjust gpt-5 reasoning parameters

* 🐛 fix: Add ExtendParamsTypeSchema and AiModelSettingsSchema for enhanced model settings

* 🐛 fix: Add ModelSearchImplementTypeSchema and update AiModelSettingsSchema for enhanced model configuration

* delete gemini-2.5-flash-image-preview model

* Add GLM-4.7 model to volcengine and remove deprecated GLM-4 32B 0414 model from wenxin

* ✨ feat: 添加 MiniMax-M2.1 和 GLM-4.7-Flash 模型到模型库

* ✨ feat: 更新 Zhipu 模型库，添加 GLM-4.7-FlashX 模型并移除 GLM-4.5-Flash 模型

* test: add extendParams mapping for gpt-5.x reasoning models in VercelAIGatewayAI

* remove deprecated DeepSeek R1 model from nvidiaChatModels

* i18n: 更新 MiniMax-M2.1 模型描述为英文

Author: sxjeru

packages/model-bank/src/aiModels/cerebras.ts

@@ -1,27 +1,6 @@
 import { AIChatModelCard } from '../types/aiModel';
 
 const cerebrasModels: AIChatModelCard[] = [
-  {
-    abilities: {
-      functionCall: true,
-      reasoning: true,
-      structuredOutput: true,
-    },
-    contextWindowTokens: 131_072,
-    description:
-      'Performs well on coding and reasoning tasks, supports streaming and tool calls, and fits agentic coding and complex reasoning.',
-    displayName: 'GLM-4.6',
-    enabled: true,
-    id: 'zai-glm-4.6',
-    maxOutput: 40_000,
-    pricing: {
-      units: [
-        { name: 'textInput', rate: 2.25, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'textOutput', rate: 2.75, strategy: 'fixed', unit: 'millionTokens' },
-      ],
-    },
-    type: 'chat',
-  },
   {
     abilities: {
       functionCall: true,
@@ -96,7 +75,8 @@ const cerebrasModels: AIChatModelCard[] = [
       functionCall: true,
     },
     contextWindowTokens: 32_768,
-    description: 'Llama 3.1 8B: a small, low-latency Llama variant for lightweight online inference and chat.',
+    description:
+      'Llama 3.1 8B: a small, low-latency Llama variant for lightweight online inference and chat.',
     displayName: 'Llama 3.1 8B',
     id: 'llama3.1-8b',
     pricing: {

packages/model-bank/src/aiModels/google.ts

@@ -1,4 +1,4 @@
-import { CHAT_MODEL_IMAGE_GENERATION_PARAMS, ModelParamsSchema } from '../standard-parameters';
+import { ModelParamsSchema } from '../standard-parameters';
 import { AIChatModelCard, AIImageModelCard } from '../types';
 
 /**
@@ -485,32 +485,6 @@ const googleChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
-  {
-    abilities: {
-      imageOutput: true,
-      vision: true,
-    },
-    contextWindowTokens: 32_768 + 8192,
-    description:
-      'Nano Banana is Google’s newest, fastest, and most efficient native multimodal model, enabling conversational image generation and editing.',
-    displayName: 'Nano Banana (Preview)',
-    id: 'gemini-2.5-flash-image-preview',
-    maxOutput: 8192,
-    pricing: {
-      approximatePricePerImage: 0.039,
-      units: [
-        { name: 'textInput', rate: 0.3, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'imageInput', rate: 0.3, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'textOutput', rate: 2.5, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'imageOutput', rate: 30, strategy: 'fixed', unit: 'millionTokens' },
-      ],
-    },
-    releasedAt: '2025-08-26',
-    settings: {
-      extendParams: ['imageAspectRatio'],
-    },
-    type: 'chat',
-  },
   {
     abilities: {
       functionCall: true,
@@ -967,23 +941,6 @@ const googleImageModels: AIImageModelCard[] = [
       ],
     },
   },
-  {
-    displayName: 'Nano Banana (Preview)',
-    id: 'gemini-2.5-flash-image-preview:image',
-    type: 'image',
-    description:
-      'Nano Banana is Google’s newest, fastest, and most efficient native multimodal model, enabling conversational image generation and editing.',
-    releasedAt: '2025-08-26',
-    parameters: CHAT_MODEL_IMAGE_GENERATION_PARAMS,
-    pricing: {
-      approximatePricePerImage: 0.039,
-      units: [
-        { name: 'textInput', rate: 0.3, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'textOutput', rate: 2.5, strategy: 'fixed', unit: 'millionTokens' },
-        { name: 'imageOutput', rate: 30, strategy: 'fixed', unit: 'millionTokens' },
-      ],
-    },
-  },
   {
     displayName: 'Imagen 4',
     id: 'imagen-4.0-generate-001',

packages/model-bank/src/aiModels/nvidia.ts

@@ -79,7 +79,8 @@ const nvidiaChatModels: AIChatModelCard[] = [
       vision: true,
     },
     contextWindowTokens: 128_000,
-    description: 'A frontier vision-language model that excels at high-quality reasoning from images.',
+    description:
+      'A frontier vision-language model that excels at high-quality reasoning from images.',
     displayName: 'Llama 3.2 11B Vision Instruct',
     id: 'meta/llama-3.2-11b-vision-instruct',
     type: 'chat',
@@ -89,7 +90,8 @@ const nvidiaChatModels: AIChatModelCard[] = [
       vision: true,
     },
     contextWindowTokens: 128_000,
-    description: 'A frontier vision-language model that excels at high-quality reasoning from images.',
+    description:
+      'A frontier vision-language model that excels at high-quality reasoning from images.',
     displayName: 'Llama 3.2 90B Vision Instruct',
     id: 'meta/llama-3.2-90b-vision-instruct',
     type: 'chat',
@@ -151,41 +153,35 @@ const nvidiaChatModels: AIChatModelCard[] = [
   },
   {
     contextWindowTokens: 8192,
-    description: 'A frontier text generation model strong in understanding, transformation, and code generation.',
+    description:
+      'A frontier text generation model strong in understanding, transformation, and code generation.',
     displayName: 'Gemma 2 9B Instruct',
     id: 'google/gemma-2-9b-it',
     type: 'chat',
   },
   {
     contextWindowTokens: 8192,
-    description: 'A frontier text generation model strong in understanding, transformation, and code generation.',
+    description:
+      'A frontier text generation model strong in understanding, transformation, and code generation.',
     displayName: 'Gemma 2 27B Instruct',
     id: 'google/gemma-2-27b-it',
     type: 'chat',
   },
-  {
-    abilities: {
-      reasoning: true,
-    },
-    contextWindowTokens: 128_000,
-    description: 'A state-of-the-art efficient LLM strong in reasoning, math, and programming.',
-    displayName: 'DeepSeek R1',
-    id: 'deepseek-ai/deepseek-r1',
-    type: 'chat',
-  },
   {
     abilities: {
       functionCall: true,
     },
     contextWindowTokens: 32_768,
-    description: 'A bilingual LLM for Chinese and English across language, coding, math, and reasoning.',
+    description:
+      'A bilingual LLM for Chinese and English across language, coding, math, and reasoning.',
     displayName: 'Qwen2.5 7B Instruct',
     id: 'qwen/qwen2.5-7b-instruct',
     type: 'chat',
   },
   {
     contextWindowTokens: 32_768,
-    description: 'A strong mid-sized code model with 32K context, excelling at multilingual programming.',
+    description:
+      'A strong mid-sized code model with 32K context, excelling at multilingual programming.',
     displayName: 'Qwen2.5 Coder 7B Instruct',
     id: 'qwen/qwen2.5-coder-7b-instruct',
     type: 'chat',

packages/model-bank/src/aiModels/siliconcloud.ts

@@ -2,6 +2,26 @@ import { AIChatModelCard, AIImageModelCard } from '../types/aiModel';
 
 // https://siliconflow.cn/zh-cn/models
 const siliconcloudChatModels: AIChatModelCard[] = [
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+    },
+    contextWindowTokens: 192_000,
+    description:
+      'MiniMax-M2.1 is an open-source large language model optimized for agent capabilities, excelling in programming, tool usage, instruction following, and long-term planning. The model supports multilingual software development and complex multi-step workflow execution, achieving a score of 74.0 on SWE-bench Verified and surpassing Claude Sonnet 4.5 in multilingual scenarios.',
+    displayName: 'MiniMax-M2.1 (Pro)',
+    id: 'Pro/MiniMaxAI/MiniMax-M2.1',
+    pricing: {
+      currency: 'CNY',
+      units: [
+        { name: 'textInput', rate: 2.1, strategy: 'fixed', unit: 'millionTokens' },
+        { name: 'textOutput', rate: 8.4, strategy: 'fixed', unit: 'millionTokens' },
+      ],
+    },
+    releasedAt: '2025-12-23',
+    type: 'chat',
+  },
   {
     abilities: {
       functionCall: true,

packages/model-bank/src/aiModels/volcengine.ts

@@ -117,6 +117,75 @@ const doubaoChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+    },
+    config: {
+      deploymentName: 'glm-4-7-251222',
+    },
+    contextWindowTokens: 200_000,
+    description:
+      'GLM-4.7 is the latest flagship model from Zhipu AI. GLM-4.7 enhances coding capabilities, long-term task planning, and tool collaboration for Agentic Coding scenarios, achieving leading performance among open-source models in multiple public benchmarks. General capabilities are improved, with more concise and natural responses, and more immersive writing. In complex agent tasks, instruction following is stronger during tool calls, and the aesthetics of Artifacts and Agentic Coding frontend, as well as long-term task completion efficiency, are further enhanced. • Stronger programming capabilities: Significantly improved multi-language coding and terminal agent performance; GLM-4.7 can now implement "think first, then act" mechanisms in programming frameworks like Claude Code, Kilo Code, TRAE, Cline, and Roo Code, with more stable performance on complex tasks. • Frontend aesthetics improvement: GLM-4.7 shows significant progress in frontend generation quality, capable of generating websites, PPTs, and posters with better visual appeal. • Stronger tool calling capabilities: GLM-4.7 enhances tool calling abilities, scoring 67 in BrowseComp web task evaluation; achieving 84.7 in τ²-Bench interactive tool calling evaluation, surpassing Claude Sonnet 4.5 as the open-source SOTA. • Reasoning capability improvement: Significantly enhanced math and reasoning abilities, scoring 42.8% in the HLE ("Humanity\'s Last Exam") benchmark, a 41% improvement over GLM-4.6, surpassing GPT-5.1. • General capability enhancement: GLM-4.7 conversations are more concise, intelligent, and humane; writing and role-playing are more literary and immersive.',
+    displayName: 'GLM-4.7',
+    id: 'glm-4-7',
+    maxOutput: 128_000,
+    pricing: {
+      currency: 'CNY',
+      units: [
+        {
+          lookup: {
+            prices: {
+              '[0, 0.032]_[0, 0.0002]': 2,
+              '[0, 0.032]_[0.0002, infinity]': 3,
+              '[0.032, 0.2]_[0, infinity]': 4,
+            },
+            pricingParams: ['textInputRange', 'textOutputRange'],
+          },
+          name: 'textInput',
+          strategy: 'lookup',
+          unit: 'millionTokens',
+        },
+        {
+          lookup: {
+            prices: {
+              '[0, 0.032]_[0, 0.0002]': 8,
+              '[0, 0.032]_[0.0002, infinity]': 14,
+              '[0.032, 0.2]_[0, infinity]': 16,
+            },
+            pricingParams: ['textInputRange', 'textOutputRange'],
+          },
+          name: 'textOutput',
+          strategy: 'lookup',
+          unit: 'millionTokens',
+        },
+        {
+          lookup: {
+            prices: {
+              '[0, 0.032]_[0, 0.0002]': 0.4,
+              '[0, 0.032]_[0.0002, infinity]': 0.6,
+              '[0.032, 0.2]_[0, infinity]': 0.8,
+            },
+            pricingParams: ['textInputRange', 'textOutputRange'],
+          },
+          name: 'textInput_cacheRead',
+          strategy: 'lookup',
+          unit: 'millionTokens',
+        },
+        {
+          lookup: { prices: { '1h': 0.017 }, pricingParams: ['ttl'] },
+          name: 'textInput_cacheWrite',
+          strategy: 'lookup',
+          unit: 'millionTokens',
+        },
+      ],
+    },
+    settings: {
+      extendParams: ['enableReasoning'],
+    },
+    type: 'chat',
+  },
   {
     abilities: {
       functionCall: true,