@vellumai/assistant 0.8.2 → 0.8.4

package/src/providers/model-catalog.ts

@@ -45,6 +45,13 @@ export interface CatalogModel {
   supportsVision?: boolean;
   supportsToolUse?: boolean;
   pricing?: CatalogModelPricing;
+  /**
+   * Upper bound for `reasoning_effort` accepted by this model's upstream API.
+   * Used by providers (e.g. Fireworks) to clamp Vellum's `xhigh`/`max` tiers
+   * down to whatever the model documents. Omit to inherit the provider
+   * default.
+   */
+  maxEffort?: "high" | "xhigh" | "max";
   /** When set, this model is only visible when the named feature flag is enabled. */
   featureFlag?: string;
 }
@@ -188,6 +195,38 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
           cacheReadPer1mTokens: 0.3,
         },
       },
+      {
+        id: "claude-sonnet-4-5-20250929",
+        displayName: "Claude Sonnet 4.5",
+        contextWindowTokens: 200000,
+        maxOutputTokens: 64000,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: {
+          inputPer1mTokens: 3,
+          outputPer1mTokens: 15,
+          cacheWritePer1mTokens: 3.75,
+          cacheReadPer1mTokens: 0.3,
+        },
+      },
+      {
+        id: "claude-opus-4-5-20251101",
+        displayName: "Claude Opus 4.5",
+        contextWindowTokens: 200000,
+        maxOutputTokens: 64000,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: {
+          inputPer1mTokens: 5,
+          outputPer1mTokens: 25,
+          cacheWritePer1mTokens: 6.25,
+          cacheReadPer1mTokens: 0.5,
+        },
+      },
       {
         id: "claude-haiku-4-5-20251001",
         displayName: "Claude Haiku 4.5",
@@ -361,6 +400,21 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
       linkLabel: "Open Google AI Studio",
     },
     models: [
+      {
+        id: "gemini-3.5-flash",
+        displayName: "Gemini 3.5 Flash",
+        contextWindowTokens: 1048576,
+        maxOutputTokens: 65536,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: {
+          inputPer1mTokens: 1.5,
+          outputPer1mTokens: 9.0,
+          cacheReadPer1mTokens: 0.15,
+        },
+      },
       {
         id: "gemini-3.1-pro-preview",
         displayName: "Gemini 3.1 Pro Preview",
@@ -548,6 +602,22 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
       linkLabel: "Open Fireworks Dashboard",
     },
     models: [
+      {
+        id: "accounts/fireworks/models/kimi-k2p6",
+        displayName: "Kimi K2.6",
+        contextWindowTokens: 262144,
+        maxOutputTokens: 32768,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        maxEffort: "high",
+        pricing: {
+          inputPer1mTokens: 0.95,
+          outputPer1mTokens: 4.0,
+          cacheReadPer1mTokens: 0.16,
+        },
+      },
       {
         id: "accounts/fireworks/models/kimi-k2p5",
         displayName: "Kimi K2.5",
@@ -562,6 +632,40 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
           outputPer1mTokens: 2.5,
         },
       },
+      {
+        id: "accounts/fireworks/models/minimax-m2p7",
+        displayName: "MiniMax M2.7",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 25000,
+        supportsThinking: false,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.3, outputPer1mTokens: 1.2 },
+      },
+      {
+        id: "accounts/fireworks/models/minimax-m2p5",
+        displayName: "MiniMax M2.5",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 25000,
+        supportsThinking: false,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.3, outputPer1mTokens: 1.2 },
+      },
+      {
+        id: "accounts/fireworks/models/deepseek-v4-pro",
+        displayName: "DeepSeek V4 Pro",
+        contextWindowTokens: 1040000,
+        maxOutputTokens: 131072,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        maxEffort: "max",
+        pricing: { inputPer1mTokens: 1.74, outputPer1mTokens: 3.48 },
+      },
     ],
     defaultModel: "accounts/fireworks/models/kimi-k2p5",
     apiKeyUrl: "https://fireworks.ai/account/api-keys",
@@ -635,6 +739,38 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
           cacheReadPer1mTokens: 0.3,
         },
       },
+      {
+        id: "anthropic/claude-sonnet-4.5",
+        displayName: "Claude Sonnet 4.5",
+        contextWindowTokens: 200000,
+        maxOutputTokens: 64000,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: {
+          inputPer1mTokens: 3,
+          outputPer1mTokens: 15,
+          cacheWritePer1mTokens: 3.75,
+          cacheReadPer1mTokens: 0.3,
+        },
+      },
+      {
+        id: "anthropic/claude-opus-4.5",
+        displayName: "Claude Opus 4.5",
+        contextWindowTokens: 200000,
+        maxOutputTokens: 64000,
+        supportsThinking: true,
+        supportsCaching: true,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: {
+          inputPer1mTokens: 5,
+          outputPer1mTokens: 25,
+          cacheWritePer1mTokens: 6.25,
+          cacheReadPer1mTokens: 0.5,
+        },
+      },
       {
         id: "anthropic/claude-haiku-4.5",
         displayName: "Claude Haiku 4.5",
@@ -663,6 +799,17 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
         supportsToolUse: true,
         pricing: { inputPer1mTokens: 3, outputPer1mTokens: 15 },
       },
+      {
+        id: "x-ai/grok-4.3",
+        displayName: "Grok 4.3",
+        contextWindowTokens: 1000000,
+        maxOutputTokens: 16000,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: true,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 1.25, outputPer1mTokens: 2.5 },
+      },
       {
         id: "x-ai/grok-4",
         displayName: "Grok 4",
@@ -798,6 +945,84 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
         supportsToolUse: true,
         pricing: { inputPer1mTokens: 0.6, outputPer1mTokens: 2.5 },
       },
+      // MiniMax
+      {
+        id: "minimax/minimax-m2.7",
+        displayName: "MiniMax M2.7",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 131072,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.279, outputPer1mTokens: 1.2 },
+      },
+      {
+        id: "minimax/minimax-m2.5",
+        displayName: "MiniMax M2.5",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 196608,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.15, outputPer1mTokens: 1.15 },
+      },
+      {
+        id: "minimax/minimax-m2.1",
+        displayName: "MiniMax M2.1",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 196608,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.29, outputPer1mTokens: 0.95 },
+      },
+      {
+        id: "minimax/minimax-m2",
+        displayName: "MiniMax M2",
+        contextWindowTokens: 196608,
+        maxOutputTokens: 196608,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.255, outputPer1mTokens: 1.0 },
+      },
+      {
+        id: "minimax/minimax-m2-her",
+        displayName: "MiniMax M2-her",
+        contextWindowTokens: 65536,
+        maxOutputTokens: 2048,
+        supportsThinking: false,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: false,
+        pricing: { inputPer1mTokens: 0.3, outputPer1mTokens: 1.2 },
+      },
+      {
+        id: "minimax/minimax-m1",
+        displayName: "MiniMax M1",
+        contextWindowTokens: 1000000,
+        maxOutputTokens: 40000,
+        supportsThinking: true,
+        supportsCaching: false,
+        supportsVision: false,
+        supportsToolUse: true,
+        pricing: { inputPer1mTokens: 0.4, outputPer1mTokens: 2.2 },
+      },
+      {
+        id: "minimax/minimax-01",
+        displayName: "MiniMax-01",
+        contextWindowTokens: 1000000,
+        maxOutputTokens: 1000000,
+        supportsThinking: false,
+        supportsCaching: false,
+        supportsVision: true,
+        supportsToolUse: false,
+        pricing: { inputPer1mTokens: 0.2, outputPer1mTokens: 1.1 },
+      },
       // Mistral
       {
         id: "mistralai/mistral-medium-3",
@@ -867,258 +1092,35 @@ const RAW_PROVIDER_CATALOG: ProviderCatalogEntry[] = [
         supportsToolUse: true,
         pricing: { inputPer1mTokens: 0.8, outputPer1mTokens: 3.2 },
       },
-    ],
-    defaultModel: "x-ai/grok-4.20-beta",
-    apiKeyUrl: "https://openrouter.ai/keys",
-    apiKeyPlaceholder: "sk-or-v1-...",
-  },
-  {
-    id: "zai",
-    displayName: "z.ai",
-    featureFlag: "provider-zai",
-    subtitle: "GLM models from z.ai (Zhipu AI). Requires a z.ai API key.",
-    setupMode: "api-key",
-    setupHint: "Enter your z.ai API key to enable GLM models.",
-    envVar: "ZAI_API_KEY",
-    credentialsGuide: {
-      description:
-        "Sign in to the z.ai open platform, navigate to API Keys, and create a new key.",
-      url: "https://open.z.ai/open/api/paas/v4/apiKey",
-      linkLabel: "Open z.ai Platform",
-    },
-    models: [
-      {
-        id: "glm-5.1",
-        displayName: "GLM-5.1",
-        contextWindowTokens: 200000,
-        maxOutputTokens: 128000,
-        supportsThinking: false,
-        supportsCaching: false,
-        supportsVision: true,
-        supportsToolUse: true,
-      },
-      {
-        id: "glm-5",
-        displayName: "GLM-5",
-        contextWindowTokens: 200000,
-        maxOutputTokens: 128000,
-        supportsThinking: false,
-        supportsCaching: false,
-        supportsVision: true,
-        supportsToolUse: true,
-      },
-      {
-        id: "glm-5-turbo",
-        displayName: "GLM-5 Turbo",
-        contextWindowTokens: 200000,
-        maxOutputTokens: 128000,
-        supportsThinking: false,
-        supportsCaching: false,
-        supportsVision: true,
-        supportsToolUse: true,
-      },
-      {
-        id: "glm-4.7",
-        displayName: "GLM-4.7",
-        contextWindowTokens: 200000,
-        maxOutputTokens: 128000,
-        supportsThinking: false,
-        supportsCaching: false,
-        supportsVision: true,
-        supportsToolUse: true,
-      },
+      // Owl (OpenRouter first-party)
       {
-        id: "glm-4.7-flash",
-        displayName: "GLM-4.7 Flash",
-        contextWindowTokens: 200000,
-        maxOutputTokens: 128000,
+        id: "openrouter/owl-alpha",
+        displayName: "Owl Alpha",
+        contextWindowTokens: 1048576,
+        maxOutputTokens: 262144,
         supportsThinking: false,
         supportsCaching: false,
-        supportsVision: true,
-        supportsToolUse: true,
-      },
-    ],
-    defaultModel: "glm-5.1",
-    apiKeyPlaceholder: "your-api-key",
-  },
-  {
-    id: "deepseek",
-    displayName: "DeepSeek",
-    subtitle:
-      "DeepSeek reasoning and chat models. Requires a DeepSeek API key.",
-    setupMode: "api-key",
-    setupHint: "Enter your DeepSeek API key to enable DeepSeek models.",
-    envVar: "DEEPSEEK_API_KEY",
-    featureFlag: "provider-deepseek",
-    credentialsGuide: {
-      description:
-        "Sign in to the DeepSeek platform, navigate to API Keys, and create a new key.",
-      url: "https://platform.deepseek.com/api_keys",
-      linkLabel: "Open DeepSeek Platform",
-    },
-    models: [
-      {
-        id: "deepseek-v4-pro",
-        displayName: "DeepSeek V4 Pro",
-        contextWindowTokens: 1000000,
-        maxOutputTokens: 384000,
-        supportsThinking: true,
-        supportsCaching: false,
         supportsVision: false,
         supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 1.74,
-          outputPer1mTokens: 3.48,
-        },
-      },
-      {
-        id: "deepseek-v4-flash",
-        displayName: "DeepSeek V4 Flash",
-        contextWindowTokens: 1000000,
-        maxOutputTokens: 384000,
-        supportsThinking: true,
-        supportsCaching: false,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.14,
-          outputPer1mTokens: 0.28,
-        },
+        pricing: { inputPer1mTokens: 0, outputPer1mTokens: 0 },
       },
     ],
-    defaultModel: "deepseek-v4-pro",
-    apiKeyPlaceholder: "sk-...",
+    defaultModel: "x-ai/grok-4.20-beta",
+    apiKeyUrl: "https://openrouter.ai/keys",
+    apiKeyPlaceholder: "sk-or-v1-...",
   },
   {
-    id: "minimax",
-    displayName: "MiniMax",
-    subtitle: "MiniMax models. Requires a MiniMax API key.",
-    featureFlag: "provider-minimax",
+    id: "openai-compatible",
+    displayName: "OpenAI-compatible",
+    subtitle:
+      "Bring your own OpenAI-compatible endpoint (vLLM, LMStudio, Groq, Together, etc.).",
     setupMode: "api-key",
-    setupHint: "Enter your MiniMax API key to enable MiniMax models.",
-    envVar: "MINIMAX_API_KEY",
-    credentialsGuide: {
-      description:
-        "Sign in to the MiniMax platform, navigate to API Keys, and create a new key.",
-      url: "https://platform.minimax.io/user-center/basic-information/interface-key",
-      linkLabel: "Open MiniMax Platform",
-    },
-    models: [
-      {
-        id: "MiniMax-M2.7",
-        displayName: "MiniMax M2.7",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 131072,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.3,
-          outputPer1mTokens: 1.2,
-          cacheReadPer1mTokens: 0.06,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2.7-highspeed",
-        displayName: "MiniMax M2.7 Highspeed",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 131072,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.6,
-          outputPer1mTokens: 2.4,
-          cacheReadPer1mTokens: 0.06,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2.5",
-        displayName: "MiniMax M2.5",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 196608,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.3,
-          outputPer1mTokens: 1.2,
-          cacheReadPer1mTokens: 0.03,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2.5-highspeed",
-        displayName: "MiniMax M2.5 Highspeed",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 196608,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.6,
-          outputPer1mTokens: 2.4,
-          cacheReadPer1mTokens: 0.03,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2.1",
-        displayName: "MiniMax M2.1",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 131072,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.3,
-          outputPer1mTokens: 1.2,
-          cacheReadPer1mTokens: 0.03,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2.1-highspeed",
-        displayName: "MiniMax M2.1 Highspeed",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 131072,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.6,
-          outputPer1mTokens: 2.4,
-          cacheReadPer1mTokens: 0.03,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-      {
-        id: "MiniMax-M2",
-        displayName: "MiniMax M2",
-        contextWindowTokens: 204800,
-        maxOutputTokens: 131072,
-        supportsThinking: true,
-        supportsCaching: true,
-        supportsVision: false,
-        supportsToolUse: true,
-        pricing: {
-          inputPer1mTokens: 0.3,
-          outputPer1mTokens: 1.2,
-          cacheReadPer1mTokens: 0.03,
-          cacheWritePer1mTokens: 0.375,
-        },
-      },
-    ],
-    defaultModel: "MiniMax-M2.7",
-    apiKeyPlaceholder: "eyJ...",
+    setupHint:
+      "Enter the base URL of your endpoint and at least one model identifier.",
+    apiKeyPlaceholder: "Your provider's API key",
+    featureFlag: "openai-compatible-endpoints",
+    models: [],
+    defaultModel: "",
   },
 ];
 

package/src/providers/model-intents.ts

@@ -35,10 +35,10 @@ const PROVIDER_MODEL_INTENTS: Record<string, Record<ModelIntent, string>> = {
     "vision-optimized": "llama3.2",
   },
   fireworks: {
-    balanced: "accounts/fireworks/models/kimi-k2p5",
+    balanced: "accounts/fireworks/models/kimi-k2p6",
     "latency-optimized": "accounts/fireworks/models/kimi-k2p5",
-    "quality-optimized": "accounts/fireworks/models/kimi-k2p5",
-    "vision-optimized": "accounts/fireworks/models/kimi-k2p5",
+    "quality-optimized": "accounts/fireworks/models/kimi-k2p6",
+    "vision-optimized": "accounts/fireworks/models/kimi-k2p6",
   },
   openrouter: {
     balanced: "anthropic/claude-sonnet-4.6",