@combycode/llm-sdk 1.5.1 → 1.6.0

package/dist/index.js

@@ -1829,6 +1829,31 @@ function parseRetryAfter(headers) {
   return void 0;
 }
 
+// src/llm/output-errors.ts
+var AgentRunError = class extends Error {
+  /** Machine-readable failure reason (e.g. `'invalid_final_output'`). */
+  reason;
+  constructor(reason, message, options) {
+    super(message, options);
+    this.name = "AgentRunError";
+    this.reason = reason;
+  }
+};
+var InvalidFinalOutputError = class extends AgentRunError {
+  reason = "invalid_final_output";
+  /** The raw model output that failed to parse. */
+  rawText;
+  constructor(rawText, options) {
+    super(
+      "invalid_final_output",
+      `Model final output did not match the requested schema: ${options?.cause instanceof Error ? options.cause.message : "parse failed"}`,
+      options
+    );
+    this.name = "InvalidFinalOutputError";
+    this.rawText = rawText;
+  }
+};
+
 // src/network/semaphore.ts
 var Semaphore = class {
   constructor(max) {
@@ -2843,7 +2868,7 @@ var catalog_default = {
     family: "claude-opus",
     version: "4",
     status: "legacy",
-    active: true,
+    active: false,
     deprecation: {
       date: "2026-05-14",
       source: "litellm"
@@ -3187,7 +3212,7 @@ var catalog_default = {
     family: "claude-sonnet",
     version: "4",
     status: "legacy",
-    active: true,
+    active: false,
     deprecation: {
       date: "2026-05-14",
       source: "litellm"
@@ -3354,50 +3379,35 @@ var catalog_default = {
     version: "4.5",
     status: "stable",
     active: true
-  }
-};
-
-// src/llm/providers/google/catalog.json
-var catalog_default2 = {
-  "google/gemini-2.5-pro": {
-    providerModelName: "gemini-2.5-pro",
+  },
+  "anthropic/claude-fable-5": {
+    providerModelName: "claude-fable-5",
     aliases: [
-      "gemini-2.5-pro"
+      "claude-fable-5"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 10,
-      cacheReadPerMTok: 0.125,
+      inputPerMTok: 10,
+      outputPerMTok: 50,
+      cacheReadPerMTok: 1,
       tiers: {
         batch: {
-          inputPerMTok: 0.625,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.125
-        },
-        flex: {
-          inputPerMTok: 0.625,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.125
-        },
-        priority: {
-          inputPerMTok: 2.25,
-          outputPerMTok: 18,
-          cacheReadPerMTok: 0.225
+          inputPerMTok: 5,
+          outputPerMTok: 25
         }
       }
     },
-    preferredApi: "generate",
+    preferredApi: "messages",
     supportedApis: [
-      "generate"
+      "messages"
     ],
-    contextWindow: 1048576,
-    maxOutput: 65536,
+    contextWindow: 1e6,
+    maxOutput: 128e3,
     capabilities: {
       toolUse: true,
       streaming: true,
       structuredOutput: true,
       vision: true,
-      audio: true,
+      audio: false,
       video: false,
       imageGeneration: false,
       audioGeneration: false,
@@ -3414,103 +3424,44 @@ var catalog_default2 = {
     inputModalities: [
       "text",
       "image",
-      "audio"
+      "pdf"
     ],
     outputModalities: [
       "text"
     ],
-    family: "gemini-pro",
-    version: "2.5",
+    family: "claude-fable",
+    version: "5",
     status: "stable",
     active: true
   },
-  "google/gemini-3.1-pro": {
-    providerModelName: "gemini-3.1-pro-preview",
+  "anthropic/claude-sonnet-5": {
+    providerModelName: "claude-sonnet-5",
     aliases: [
-      "gemini-3.1-pro-preview"
+      "claude-sonnet-5"
     ],
     pricing: {
       inputPerMTok: 2,
-      outputPerMTok: 12,
+      outputPerMTok: 10,
       cacheReadPerMTok: 0.2,
       tiers: {
         batch: {
           inputPerMTok: 1,
-          outputPerMTok: 6,
-          cacheReadPerMTok: 0.2
-        },
-        flex: {
-          inputPerMTok: 1,
-          outputPerMTok: 6,
-          cacheReadPerMTok: 0.2
-        },
-        priority: {
-          inputPerMTok: 3.6,
-          outputPerMTok: 21.6,
-          cacheReadPerMTok: 0.36
+          outputPerMTok: 5
         }
       }
     },
-    preferredApi: "generate",
-    supportedApis: [
-      "generate"
-    ],
-    contextWindow: 1048576,
-    maxOutput: 65536,
-    capabilities: {
-      toolUse: true,
-      streaming: true,
-      structuredOutput: true,
-      vision: true,
-      audio: true,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: false,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: true,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "chat",
-    inputModalities: [
-      "text",
-      "image",
-      "audio"
-    ],
-    outputModalities: [
-      "text"
-    ],
-    family: "gemini-pro",
-    version: "3.1",
-    status: "preview",
-    active: true
-  },
-  "google/gemini-3.1-pro-customtools": {
-    providerModelName: "gemini-3.1-pro-preview-customtools",
-    aliases: [
-      "gemini-3.1-pro-preview-customtools"
-    ],
-    pricing: {
-      inputPerMTok: 2,
-      outputPerMTok: 12,
-      cacheReadPerMTok: 0.2
-    },
-    preferredApi: "generate",
+    preferredApi: "messages",
     supportedApis: [
-      "generate"
+      "messages"
     ],
-    contextWindow: 1048576,
-    maxOutput: 65536,
+    contextWindow: 1e6,
+    maxOutput: 128e3,
     capabilities: {
       toolUse: true,
       streaming: true,
       structuredOutput: true,
       vision: true,
-      audio: true,
+      audio: false,
       video: false,
       imageGeneration: false,
       audioGeneration: false,
@@ -3527,66 +3478,49 @@ var catalog_default2 = {
     inputModalities: [
       "text",
       "image",
-      "audio"
+      "pdf"
     ],
     outputModalities: [
       "text"
     ],
-    family: "gemini-pro",
-    version: "3.1",
-    status: "preview",
+    family: "claude-sonnet",
+    version: "5",
+    status: "stable",
     active: true
   },
-  "google/gemini-2.5-flash": {
-    providerModelName: "gemini-2.5-flash",
+  "anthropic/claude-mythos-5": {
+    providerModelName: "claude-mythos-5",
     aliases: [
-      "gemini-2.5-flash"
+      "claude-mythos-5"
     ],
     pricing: {
-      inputPerMTok: 0.3,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.03,
-      audioInputPerMTok: 1,
+      inputPerMTok: 10,
+      outputPerMTok: 50,
+      cacheReadPerMTok: 1,
       tiers: {
         batch: {
-          inputPerMTok: 0.15,
-          outputPerMTok: 1.25,
-          cacheReadPerMTok: 0.03,
-          audioInputPerMTok: 0.5
-        },
-        flex: {
-          inputPerMTok: 0.15,
-          outputPerMTok: 1.25,
-          cacheReadPerMTok: 0.03,
-          audioInputPerMTok: 0.5
-        },
-        priority: {
-          inputPerMTok: 0.54,
-          outputPerMTok: 4.5,
-          cacheReadPerMTok: 0.054,
-          audioInputPerMTok: 1.8
+          inputPerMTok: 5,
+          outputPerMTok: 25
         }
       }
     },
-    preferredApi: "generate",
+    preferredApi: "messages",
     supportedApis: [
-      "generate"
+      "messages"
     ],
-    contextWindow: 1048576,
-    maxOutput: 65536,
     capabilities: {
       toolUse: true,
       streaming: true,
       structuredOutput: true,
-      vision: true,
-      audio: true,
+      vision: false,
+      audio: false,
       video: false,
       imageGeneration: false,
       audioGeneration: false,
       videoGeneration: false
     },
     reasoning: {
-      supported: true,
+      supported: false,
       automatic: false,
       effortControl: false,
       encryptedContent: false,
@@ -3594,46 +3528,45 @@ var catalog_default2 = {
     },
     type: "chat",
     inputModalities: [
-      "text",
-      "image",
-      "audio"
+      "text"
     ],
     outputModalities: [
       "text"
     ],
-    family: "gemini-flash",
-    version: "2.5",
+    family: "claude-mythos",
+    version: "5",
     status: "stable",
-    active: true
-  },
-  "google/gemini-3-flash": {
-    providerModelName: "gemini-3-flash-preview",
+    active: true,
+    availability: "limited"
+  }
+};
+
+// src/llm/providers/google/catalog.json
+var catalog_default2 = {
+  "google/gemini-2.5-pro": {
+    providerModelName: "gemini-2.5-pro",
     aliases: [
-      "gemini-3-flash-preview"
+      "gemini-2.5-pro"
     ],
     pricing: {
-      inputPerMTok: 0.5,
-      outputPerMTok: 3,
-      cacheReadPerMTok: 0.05,
-      audioInputPerMTok: 1,
+      inputPerMTok: 1.25,
+      outputPerMTok: 10,
+      cacheReadPerMTok: 0.125,
       tiers: {
         batch: {
-          inputPerMTok: 0.25,
-          outputPerMTok: 1.5,
-          cacheReadPerMTok: 0.05,
-          audioInputPerMTok: 0.5
+          inputPerMTok: 0.625,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.125
         },
         flex: {
-          inputPerMTok: 0.25,
-          outputPerMTok: 1.5,
-          cacheReadPerMTok: 0.05,
-          audioInputPerMTok: 0.5
+          inputPerMTok: 0.625,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.125
         },
         priority: {
-          inputPerMTok: 0.9,
-          outputPerMTok: 5.4,
-          cacheReadPerMTok: 0.09,
-          audioInputPerMTok: 1.8
+          inputPerMTok: 2.25,
+          outputPerMTok: 18,
+          cacheReadPerMTok: 0.225
         }
       }
     },
@@ -3670,35 +3603,286 @@ var catalog_default2 = {
     outputModalities: [
       "text"
     ],
-    family: "gemini-flash",
-    version: "3",
-    status: "preview",
+    family: "gemini-pro",
+    version: "2.5",
+    status: "stable",
     active: true
   },
-  "google/gemini-3.5-flash": {
-    providerModelName: "gemini-3.5-flash",
+  "google/gemini-3.1-pro": {
+    providerModelName: "gemini-3.1-pro-preview",
     aliases: [
-      "gemini-3.5-flash"
+      "gemini-3.1-pro-preview"
     ],
     pricing: {
-      inputPerMTok: 1.5,
-      outputPerMTok: 9,
-      cacheReadPerMTok: 0.15,
+      inputPerMTok: 2,
+      outputPerMTok: 12,
+      cacheReadPerMTok: 0.2,
       tiers: {
         batch: {
-          inputPerMTok: 0.75,
-          outputPerMTok: 4.5,
-          cacheReadPerMTok: 0.075
+          inputPerMTok: 1,
+          outputPerMTok: 6,
+          cacheReadPerMTok: 0.2
         },
         flex: {
-          inputPerMTok: 0.75,
-          outputPerMTok: 4.5,
-          cacheReadPerMTok: 0.08
+          inputPerMTok: 1,
+          outputPerMTok: 6,
+          cacheReadPerMTok: 0.2
         },
         priority: {
-          inputPerMTok: 2.7,
-          outputPerMTok: 16.2,
-          cacheReadPerMTok: 0.27
+          inputPerMTok: 3.6,
+          outputPerMTok: 21.6,
+          cacheReadPerMTok: 0.36
+        }
+      }
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 1048576,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gemini-pro",
+    version: "3.1",
+    status: "preview",
+    active: true
+  },
+  "google/gemini-3.1-pro-customtools": {
+    providerModelName: "gemini-3.1-pro-preview-customtools",
+    aliases: [
+      "gemini-3.1-pro-preview-customtools"
+    ],
+    pricing: {
+      inputPerMTok: 2,
+      outputPerMTok: 12,
+      cacheReadPerMTok: 0.2
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 1048576,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gemini-pro",
+    version: "3.1",
+    status: "preview",
+    active: true
+  },
+  "google/gemini-2.5-flash": {
+    providerModelName: "gemini-2.5-flash",
+    aliases: [
+      "gemini-2.5-flash"
+    ],
+    pricing: {
+      inputPerMTok: 0.3,
+      outputPerMTok: 2.5,
+      cacheReadPerMTok: 0.03,
+      audioInputPerMTok: 1,
+      tiers: {
+        batch: {
+          inputPerMTok: 0.15,
+          outputPerMTok: 1.25,
+          cacheReadPerMTok: 0.03,
+          audioInputPerMTok: 0.5
+        },
+        flex: {
+          inputPerMTok: 0.15,
+          outputPerMTok: 1.25,
+          cacheReadPerMTok: 0.03,
+          audioInputPerMTok: 0.5
+        },
+        priority: {
+          inputPerMTok: 0.54,
+          outputPerMTok: 4.5,
+          cacheReadPerMTok: 0.054,
+          audioInputPerMTok: 1.8
+        }
+      }
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 1048576,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gemini-flash",
+    version: "2.5",
+    status: "stable",
+    active: true
+  },
+  "google/gemini-3-flash": {
+    providerModelName: "gemini-3-flash-preview",
+    aliases: [
+      "gemini-3-flash-preview"
+    ],
+    pricing: {
+      inputPerMTok: 0.5,
+      outputPerMTok: 3,
+      cacheReadPerMTok: 0.05,
+      audioInputPerMTok: 1,
+      tiers: {
+        batch: {
+          inputPerMTok: 0.25,
+          outputPerMTok: 1.5,
+          cacheReadPerMTok: 0.05,
+          audioInputPerMTok: 0.5
+        },
+        flex: {
+          inputPerMTok: 0.25,
+          outputPerMTok: 1.5,
+          cacheReadPerMTok: 0.05,
+          audioInputPerMTok: 0.5
+        },
+        priority: {
+          inputPerMTok: 0.9,
+          outputPerMTok: 5.4,
+          cacheReadPerMTok: 0.09,
+          audioInputPerMTok: 1.8
+        }
+      }
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 1048576,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gemini-flash",
+    version: "3",
+    status: "preview",
+    active: true
+  },
+  "google/gemini-3.5-flash": {
+    providerModelName: "gemini-3.5-flash",
+    aliases: [
+      "gemini-3.5-flash"
+    ],
+    pricing: {
+      inputPerMTok: 1.5,
+      outputPerMTok: 9,
+      cacheReadPerMTok: 0.15,
+      tiers: {
+        batch: {
+          inputPerMTok: 0.75,
+          outputPerMTok: 4.5,
+          cacheReadPerMTok: 0.075
+        },
+        flex: {
+          inputPerMTok: 0.75,
+          outputPerMTok: 4.5,
+          cacheReadPerMTok: 0.08
+        },
+        priority: {
+          inputPerMTok: 2.7,
+          outputPerMTok: 16.2,
+          cacheReadPerMTok: 0.27
         }
       }
     },
@@ -4549,7 +4733,8 @@ var catalog_default2 = {
       outputPerMTok: 4.5,
       audioInputPerMTok: 3,
       audioOutputPerMTok: 12,
-      perMinute: 5e-3
+      perMinute: 5e-3,
+      perSecond: 2e-3
     },
     preferredApi: "generate",
     supportedApis: [
@@ -4598,7 +4783,9 @@ var catalog_default2 = {
     pricing: {
       inputPerMTok: 3.5,
       outputPerMTok: 21,
-      perMinute: 53e-4
+      perMinute: 53e-4,
+      audioInputPerMTok: 53e-4,
+      audioOutputPerMTok: 0.0315
     },
     preferredApi: "generate",
     supportedApis: [
@@ -5172,7 +5359,8 @@ var catalog_default2 = {
         "720p": 0.1,
         "1080p": 0.12,
         "4k": 0.3
-      }
+      },
+      perSecond: 0.1
     },
     preferredApi: "generate",
     supportedApis: [
@@ -5319,7 +5507,8 @@ var catalog_default2 = {
         "720p": 0.1,
         "1080p": 0.12,
         "4k": 0.3
-      }
+      },
+      perSecond: 0.1
     },
     preferredApi: "generate",
     supportedApis: [
@@ -5390,8 +5579,10 @@ var catalog_default2 = {
     pricing: {
       perUnit: {
         "720p": 0.05,
-        "1080p": 0.08
-      }
+        "1080p": 0.08,
+        "4k": 0
+      },
+      perSecond: 0.05
     },
     preferredApi: "generate",
     supportedApis: [
@@ -5692,6 +5883,129 @@ var catalog_default2 = {
     version: "1.6",
     status: "preview",
     active: true
+  },
+  "google/gemini-3.1-flash-lite-image": {
+    providerModelName: "gemini-3.1-flash-lite-image",
+    aliases: [
+      "gemini-3.1-flash-lite-image"
+    ],
+    pricing: {
+      inputPerMTok: 0.25,
+      outputPerMTok: 1.5,
+      perUnit: {
+        "1k": 0.0336
+      }
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 65536,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: true,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "image",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "image"
+    ],
+    family: "gemini-image",
+    version: "3.1",
+    status: "stable",
+    active: true,
+    mediaParams: {
+      aspectRatio: {
+        values: [
+          "1:1",
+          "1:4",
+          "1:8",
+          "2:3",
+          "3:2",
+          "3:4",
+          "4:1",
+          "4:3",
+          "4:5",
+          "5:4",
+          "9:16",
+          "16:9",
+          "21:9"
+        ],
+        default: "1:1"
+      },
+      imageSize: {
+        values: [
+          "512",
+          "1K",
+          "2K",
+          "4K"
+        ],
+        default: "1K"
+      }
+    }
+  },
+  "google/gemini-omni-flash": {
+    providerModelName: "gemini-omni-flash-preview",
+    aliases: [
+      "gemini-omni-flash-preview"
+    ],
+    pricing: {
+      inputPerMTok: 1.5
+    },
+    preferredApi: "generate",
+    supportedApis: [
+      "generate"
+    ],
+    contextWindow: 131072,
+    maxOutput: 65536,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "audio-chat",
+    inputModalities: [
+      "text",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gemini-audio",
+    version: "1",
+    status: "preview",
+    active: true
   }
 };
 
@@ -7291,9 +7605,10 @@ var catalog_default3 = {
       "gpt-4o-mini-2024-07-18"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 5,
-      perMinute: 3e-3
+      inputPerMTok: 0.15,
+      outputPerMTok: 0.6,
+      perMinute: 3e-3,
+      cacheReadPerMTok: 0.075
     },
     preferredApi: "responses",
     supportedApis: [
@@ -7946,8 +8261,8 @@ var catalog_default3 = {
       "gpt-realtime-2025-08-28"
     ],
     pricing: {
-      inputPerMTok: 4,
-      outputPerMTok: 24,
+      inputPerMTok: 32,
+      outputPerMTok: 64,
       cacheReadPerMTok: 0.4,
       audioInputPerMTok: 32,
       audioOutputPerMTok: 64
@@ -8046,8 +8361,8 @@ var catalog_default3 = {
       "gpt-realtime-2"
     ],
     pricing: {
-      inputPerMTok: 4,
-      outputPerMTok: 24,
+      inputPerMTok: 32,
+      outputPerMTok: 64,
       cacheReadPerMTok: 0.4,
       audioInputPerMTok: 32,
       audioOutputPerMTok: 64
@@ -8182,7 +8497,8 @@ var catalog_default3 = {
     family: "gpt-realtime",
     version: "1",
     status: "stable",
-    active: true
+    active: true,
+    contextWindow: 128e3
   },
   "openai/gpt-realtime-whisper": {
     providerModelName: "gpt-realtime-whisper",
@@ -8225,7 +8541,8 @@ var catalog_default3 = {
     family: "gpt-realtime",
     version: "1",
     status: "stable",
-    active: true
+    active: true,
+    contextWindow: 128e3
   },
   "openai/o1": {
     providerModelName: "o1",
@@ -8334,7 +8651,8 @@ var catalog_default3 = {
     },
     preferredApi: "responses",
     supportedApis: [
-      "responses"
+      "responses",
+      "completions"
     ],
     contextWindow: 2e5,
     maxOutput: 1e5,
@@ -8468,9 +8786,9 @@ var catalog_default3 = {
       "o4-mini-2025-04-16"
     ],
     pricing: {
-      inputPerMTok: 4,
-      outputPerMTok: 16,
-      cacheReadPerMTok: 1,
+      inputPerMTok: 1.1,
+      outputPerMTok: 4.4,
+      cacheReadPerMTok: 0.275,
       tiers: {
         batch: {
           inputPerMTok: 2,
@@ -8807,167 +9125,171 @@ var catalog_default3 = {
       "tts-1",
       "tts-1-1106"
     ],
-    pricing: { perMChars: 15 },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: false,
-      audio: false,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: true,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "tts",
-    inputModalities: [
-      "text"
-    ],
-    outputModalities: [
-      "audio"
-    ],
-    family: "tts",
-    version: "1",
-    status: "legacy",
-    active: true,
-    mediaParams: {
-      voice: {
-        values: [
-          "alloy",
-          "ash",
-          "ballad",
-          "coral",
-          "echo",
-          "fable",
-          "onyx",
-          "nova",
-          "sage",
-          "shimmer",
-          "verse",
-          "marin",
-          "cedar"
-        ],
-        default: "alloy"
-      },
-      format: {
-        values: [
-          "mp3",
-          "opus",
-          "aac",
-          "flac",
-          "wav",
-          "pcm"
-        ],
-        default: "mp3"
-      },
-      speed: {
-        min: 0.25,
-        max: 4,
-        default: 1
-      }
-    }
-  },
-  "openai/tts-1-hd": {
-    providerModelName: "tts-1-hd",
-    aliases: [
-      "tts-1-hd",
-      "tts-1-hd-1106"
-    ],
-    pricing: { perMChars: 30 },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: false,
-      audio: false,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: true,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "tts",
-    inputModalities: [
-      "text"
-    ],
-    outputModalities: [
-      "audio"
-    ],
-    family: "tts",
-    version: "1",
-    status: "legacy",
-    active: true,
-    mediaParams: {
-      voice: {
-        values: [
-          "alloy",
-          "ash",
-          "ballad",
-          "coral",
-          "echo",
-          "fable",
-          "onyx",
-          "nova",
-          "sage",
-          "shimmer",
-          "verse",
-          "marin",
-          "cedar"
-        ],
-        default: "alloy"
-      },
-      format: {
-        values: [
-          "mp3",
-          "opus",
-          "aac",
-          "flac",
-          "wav",
-          "pcm"
-        ],
-        default: "mp3"
-      },
-      speed: {
-        min: 0.25,
-        max: 4,
-        default: 1
-      }
-    }
-  },
-  "openai/gpt-4o-mini-tts": {
-    providerModelName: "gpt-4o-mini-tts",
-    aliases: [
-      "gpt-4o-mini-tts",
-      "gpt-4o-mini-tts-2025-03-20",
-      "gpt-4o-mini-tts-2025-12-15"
-    ],
     pricing: {
-      inputPerMTok: 2.5,
-      outputPerMTok: 10,
-      audioOutputPerMTok: 12
+      perMChars: 15
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: true,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "tts",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "audio"
+    ],
+    family: "tts",
+    version: "1",
+    status: "legacy",
+    active: true,
+    mediaParams: {
+      voice: {
+        values: [
+          "alloy",
+          "ash",
+          "ballad",
+          "coral",
+          "echo",
+          "fable",
+          "onyx",
+          "nova",
+          "sage",
+          "shimmer",
+          "verse",
+          "marin",
+          "cedar"
+        ],
+        default: "alloy"
+      },
+      format: {
+        values: [
+          "mp3",
+          "opus",
+          "aac",
+          "flac",
+          "wav",
+          "pcm"
+        ],
+        default: "mp3"
+      },
+      speed: {
+        min: 0.25,
+        max: 4,
+        default: 1
+      }
+    }
+  },
+  "openai/tts-1-hd": {
+    providerModelName: "tts-1-hd",
+    aliases: [
+      "tts-1-hd",
+      "tts-1-hd-1106"
+    ],
+    pricing: {
+      perMChars: 30
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: true,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "tts",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "audio"
+    ],
+    family: "tts",
+    version: "1",
+    status: "legacy",
+    active: true,
+    mediaParams: {
+      voice: {
+        values: [
+          "alloy",
+          "ash",
+          "ballad",
+          "coral",
+          "echo",
+          "fable",
+          "onyx",
+          "nova",
+          "sage",
+          "shimmer",
+          "verse",
+          "marin",
+          "cedar"
+        ],
+        default: "alloy"
+      },
+      format: {
+        values: [
+          "mp3",
+          "opus",
+          "aac",
+          "flac",
+          "wav",
+          "pcm"
+        ],
+        default: "mp3"
+      },
+      speed: {
+        min: 0.25,
+        max: 4,
+        default: 1
+      }
+    }
+  },
+  "openai/gpt-4o-mini-tts": {
+    providerModelName: "gpt-4o-mini-tts",
+    aliases: [
+      "gpt-4o-mini-tts",
+      "gpt-4o-mini-tts-2025-03-20",
+      "gpt-4o-mini-tts-2025-12-15"
+    ],
+    pricing: {
+      inputPerMTok: 2.5,
+      outputPerMTok: 10,
+      audioOutputPerMTok: 12
     },
     preferredApi: "responses",
     supportedApis: [
@@ -9045,7 +9367,9 @@ var catalog_default3 = {
     aliases: [
       "whisper-1"
     ],
-    pricing: { perMinute: 6e-3 },
+    pricing: {
+      perMinute: 6e-3
+    },
     preferredApi: "responses",
     supportedApis: [
       "responses",
@@ -9079,7 +9403,8 @@ var catalog_default3 = {
     family: "whisper",
     version: "1",
     status: "legacy",
-    active: true
+    active: true,
+    contextWindow: 16e3
   },
   "openai/gpt-4o-transcribe": {
     providerModelName: "gpt-4o-transcribe",
@@ -9271,6 +9596,300 @@ var catalog_default3 = {
     version: "1",
     status: "stable",
     active: true
+  },
+  "openai/gpt-realtime-2.1-mini": {
+    providerModelName: "gpt-realtime-2.1-mini",
+    aliases: [
+      "gpt-realtime-2.1-mini"
+    ],
+    pricing: {
+      inputPerMTok: 10,
+      cacheReadPerMTok: 0.3,
+      outputPerMTok: 20
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 128e3,
+    maxOutput: 4096,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: false,
+      vision: false,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "realtime",
+    inputModalities: [
+      "text",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-realtime",
+    version: "2.1",
+    status: "stable",
+    active: true
+  },
+  "openai/gpt-realtime-2.1": {
+    providerModelName: "gpt-realtime-2.1",
+    aliases: [
+      "gpt-realtime-2.1"
+    ],
+    pricing: {
+      inputPerMTok: 32,
+      cacheReadPerMTok: 0.4,
+      outputPerMTok: 64
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 128e3,
+    maxOutput: 32e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: false,
+      vision: false,
+      audio: true,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "realtime",
+    inputModalities: [
+      "text",
+      "audio"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-realtime",
+    version: "2.1",
+    status: "stable",
+    active: true
+  },
+  "openai/gpt-5.6-sol": {
+    providerModelName: "gpt-5.6-sol",
+    aliases: [
+      "gpt-5.6-sol"
+    ],
+    pricing: {
+      inputPerMTok: 5,
+      outputPerMTok: 30,
+      cacheReadPerMTok: 0.5,
+      tiers: {
+        batch: {
+          inputPerMTok: 2.5,
+          outputPerMTok: 15,
+          cacheReadPerMTok: 0.25
+        },
+        flex: {
+          inputPerMTok: 2.5,
+          outputPerMTok: 15,
+          cacheReadPerMTok: 0.25
+        },
+        priority: {
+          inputPerMTok: 10,
+          outputPerMTok: 60,
+          cacheReadPerMTok: 1
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false,
+      webSearch: true
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-sol",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openai/gpt-5.6-terra": {
+    providerModelName: "gpt-5.6-terra",
+    aliases: [
+      "gpt-5.6-terra"
+    ],
+    pricing: {
+      inputPerMTok: 2.5,
+      outputPerMTok: 15,
+      cacheReadPerMTok: 0.25,
+      tiers: {
+        batch: {
+          inputPerMTok: 1.25,
+          outputPerMTok: 7.5,
+          cacheReadPerMTok: 0.125
+        },
+        flex: {
+          inputPerMTok: 1.25,
+          outputPerMTok: 7.5,
+          cacheReadPerMTok: 0.125
+        },
+        priority: {
+          inputPerMTok: 5,
+          outputPerMTok: 30,
+          cacheReadPerMTok: 0.5
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false,
+      webSearch: true
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-terra",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openai/gpt-5.6-luna": {
+    providerModelName: "gpt-5.6-luna",
+    aliases: [
+      "gpt-5.6-luna"
+    ],
+    pricing: {
+      inputPerMTok: 1,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.1,
+      tiers: {
+        batch: {
+          inputPerMTok: 0.5,
+          outputPerMTok: 3,
+          cacheReadPerMTok: 0.05
+        },
+        flex: {
+          inputPerMTok: 0.5,
+          outputPerMTok: 3,
+          cacheReadPerMTok: 0.05
+        },
+        priority: {
+          inputPerMTok: 2,
+          outputPerMTok: 12,
+          cacheReadPerMTok: 0.2
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false,
+      webSearch: true
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-luna",
+    version: "5.6",
+    status: "stable",
+    active: true
   }
 };
 
@@ -9378,7 +9997,7 @@ var catalog_default4 = {
     supportedApis: [
       "completions"
     ],
-    contextWindow: 128e3,
+    contextWindow: 1e6,
     capabilities: {
       toolUse: false,
       streaming: true,
@@ -9737,7 +10356,7 @@ var catalog_default4 = {
     capabilities: {
       toolUse: false,
       streaming: true,
-      structuredOutput: false,
+      structuredOutput: true,
       vision: false,
       audio: false,
       video: false,
@@ -12337,7 +12956,7 @@ var catalog_default4 = {
       "openai/gpt-oss-120b:free"
     ],
     pricing: {
-      inputPerMTok: 0.039,
+      inputPerMTok: 0.036,
       outputPerMTok: 0.18
     },
     preferredApi: "completions",
@@ -13388,7 +14007,8 @@ var catalog_default4 = {
   "openrouter/gemini-3-pro-image": {
     providerModelName: "google/gemini-3-pro-image-preview",
     aliases: [
-      "google/gemini-3-pro-image-preview"
+      "google/gemini-3-pro-image-preview",
+      "google/gemini-3-pro-image"
     ],
     pricing: {
       inputPerMTok: 2,
@@ -13460,7 +14080,8 @@ var catalog_default4 = {
   "openrouter/gemini-3.1-flash-image": {
     providerModelName: "google/gemini-3.1-flash-image-preview",
     aliases: [
-      "google/gemini-3.1-flash-image-preview"
+      "google/gemini-3.1-flash-image-preview",
+      "google/gemini-3.1-flash-image"
     ],
     pricing: {
       inputPerMTok: 0.5,
@@ -13993,8 +14614,8 @@ var catalog_default4 = {
       "x-ai/grok-4.20-multi-agent"
     ],
     pricing: {
-      inputPerMTok: 2,
-      outputPerMTok: 6,
+      inputPerMTok: 1.25,
+      outputPerMTok: 2.5,
       cacheReadPerMTok: 0.2
     },
     preferredApi: "completions",
@@ -14176,8 +14797,8 @@ var catalog_default4 = {
       "deepseek/deepseek-chat-v3-0324"
     ],
     pricing: {
-      inputPerMTok: 0.2,
-      outputPerMTok: 0.77,
+      inputPerMTok: 0.24,
+      outputPerMTok: 0.9,
       cacheReadPerMTok: 0.135
     },
     preferredApi: "completions",
@@ -14270,8 +14891,9 @@ var catalog_default4 = {
       "deepseek/deepseek-v3.2-exp"
     ],
     pricing: {
-      inputPerMTok: 0.2288,
-      outputPerMTok: 0.3432
+      inputPerMTok: 0.2145,
+      outputPerMTok: 0.32175,
+      cacheReadPerMTok: 0.02145
     },
     preferredApi: "completions",
     supportedApis: [
@@ -14315,9 +14937,9 @@ var catalog_default4 = {
       "deepseek/deepseek-v4-flash"
     ],
     pricing: {
-      inputPerMTok: 0.098,
-      outputPerMTok: 0.196,
-      cacheReadPerMTok: 0.02
+      inputPerMTok: 0.077,
+      outputPerMTok: 0.154,
+      cacheReadPerMTok: 0.0154
     },
     preferredApi: "completions",
     supportedApis: [
@@ -14679,8 +15301,8 @@ var catalog_default4 = {
       "qwen/qwen3-8b"
     ],
     pricing: {
-      inputPerMTok: 0.05,
-      outputPerMTok: 0.4,
+      inputPerMTok: 0.117,
+      outputPerMTok: 0.455,
       cacheReadPerMTok: 0.05
     },
     preferredApi: "completions",
@@ -15141,14 +15763,15 @@ var catalog_default4 = {
       "qwen/qwen3.5-397b-a17b"
     ],
     pricing: {
-      inputPerMTok: 0.39,
-      outputPerMTok: 2.34
+      inputPerMTok: 0.385,
+      outputPerMTok: 2.45,
+      cacheReadPerMTok: 0.111
     },
     preferredApi: "completions",
     supportedApis: [
       "completions"
     ],
-    contextWindow: 262144,
+    contextWindow: 256e3,
     maxOutput: 65536,
     capabilities: {
       toolUse: true,
@@ -15187,8 +15810,9 @@ var catalog_default4 = {
       "qwen/qwen3.6-27b"
     ],
     pricing: {
-      inputPerMTok: 0.2885,
-      outputPerMTok: 3.17
+      inputPerMTok: 0.285,
+      outputPerMTok: 2.4,
+      cacheReadPerMTok: 0.15
     },
     preferredApi: "completions",
     supportedApis: [
@@ -15233,7 +15857,7 @@ var catalog_default4 = {
       "qwen/qwen3.6-35b-a3b"
     ],
     pricing: {
-      inputPerMTok: 0.15,
+      inputPerMTok: 0.14,
       outputPerMTok: 1,
       cacheReadPerMTok: 0.05
     },
@@ -15838,8 +16462,8 @@ var catalog_default4 = {
       "qwen/qwen3-vl-8b-thinking"
     ],
     pricing: {
-      inputPerMTok: 0.08,
-      outputPerMTok: 0.5
+      inputPerMTok: 0.117,
+      outputPerMTok: 0.455
     },
     preferredApi: "completions",
     supportedApis: [
@@ -16254,8 +16878,8 @@ var catalog_default4 = {
       "meta-llama/llama-3.2-3b-instruct:free"
     ],
     pricing: {
-      inputPerMTok: 0.0509,
-      outputPerMTok: 0.335
+      inputPerMTok: 0.05,
+      outputPerMTok: 0.33
     },
     preferredApi: "completions",
     supportedApis: [
@@ -16337,7 +16961,11 @@ var catalog_default4 = {
     family: "llama",
     version: "3.2",
     status: "legacy",
-    active: true
+    active: true,
+    deprecation: {
+      shutdownDate: "2026-07-17",
+      source: "openrouter"
+    }
   },
   "openrouter/llama-3.3-70b": {
     providerModelName: "meta-llama/llama-3.3-70b-instruct",
@@ -16515,7 +17143,8 @@ var catalog_default4 = {
     family: "llama-guard",
     version: "3",
     status: "stable",
-    active: true
+    active: true,
+    contextWindow: 163840
   },
   "openrouter/llama-guard-4-12b": {
     providerModelName: "meta-llama/llama-guard-4-12b",
@@ -17219,9 +17848,9 @@ var catalog_default4 = {
       "moonshotai/kimi-k2.6"
     ],
     pricing: {
-      inputPerMTok: 0.68,
+      inputPerMTok: 0.66,
       outputPerMTok: 3.41,
-      cacheReadPerMTok: 0.34
+      cacheReadPerMTok: 0.15
     },
     preferredApi: "completions",
     supportedApis: [
@@ -17308,7 +17937,7 @@ var catalog_default4 = {
     status: "legacy",
     active: true,
     deprecation: {
-      shutdownDate: "2026-06-19",
+      shutdownDate: "2026-12-31",
       source: "openrouter"
     }
   },
@@ -17459,9 +18088,9 @@ var catalog_default4 = {
       "z-ai/glm-5.1"
     ],
     pricing: {
-      inputPerMTok: 0.98,
-      outputPerMTok: 3.08,
-      cacheReadPerMTok: 0.182
+      inputPerMTok: 0.966,
+      outputPerMTok: 3.036,
+      cacheReadPerMTok: 0.1794
     },
     preferredApi: "completions",
     supportedApis: [
@@ -17598,7 +18227,7 @@ var catalog_default4 = {
     ],
     pricing: {
       inputPerMTok: 0.255,
-      outputPerMTok: 1,
+      outputPerMTok: 1.02,
       cacheReadPerMTok: 0.03
     },
     preferredApi: "completions",
@@ -17689,8 +18318,8 @@ var catalog_default4 = {
       "minimax/minimax-m2.7"
     ],
     pricing: {
-      inputPerMTok: 0.25,
-      outputPerMTok: 1,
+      inputPerMTok: 0.24,
+      outputPerMTok: 0.96,
       cacheReadPerMTok: 0.05
     },
     preferredApi: "completions",
@@ -17909,7 +18538,11 @@ var catalog_default4 = {
     family: "nemotron",
     version: "1.5",
     status: "stable",
-    active: true
+    active: true,
+    deprecation: {
+      shutdownDate: "2026-07-17",
+      source: "openrouter"
+    }
   },
   "openrouter/nemotron-3-nano-30b-a3b": {
     providerModelName: "nvidia/nemotron-3-nano-30b-a3b",
@@ -17965,7 +18598,7 @@ var catalog_default4 = {
       "nvidia/nemotron-3-super-120b-a12b:free"
     ],
     pricing: {
-      inputPerMTok: 0.09,
+      inputPerMTok: 0.08,
       outputPerMTok: 0.45
     },
     preferredApi: "completions",
@@ -18011,8 +18644,8 @@ var catalog_default4 = {
     ],
     pricing: {
       inputPerMTok: 0.5,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.15
+      outputPerMTok: 2.2,
+      cacheReadPerMTok: 0.1
     },
     preferredApi: "completions",
     supportedApis: [
@@ -18512,7 +19145,7 @@ var catalog_default4 = {
       "microsoft/phi-4"
     ],
     pricing: {
-      inputPerMTok: 0.065,
+      inputPerMTok: 0.07,
       outputPerMTok: 0.14
     },
     preferredApi: "completions",
@@ -19379,8 +20012,8 @@ var catalog_default4 = {
       "arcee-ai/trinity-large-thinking"
     ],
     pricing: {
-      inputPerMTok: 0.22,
-      outputPerMTok: 0.85,
+      inputPerMTok: 0.25,
+      outputPerMTok: 0.8,
       cacheReadPerMTok: 0.06
     },
     preferredApi: "completions",
@@ -20329,7 +20962,7 @@ var catalog_default4 = {
       "stepfun/step-3.5-flash"
     ],
     pricing: {
-      inputPerMTok: 0.09,
+      inputPerMTok: 0.1,
       outputPerMTok: 0.3,
       cacheReadPerMTok: 0.02
     },
@@ -20464,7 +21097,9 @@ var catalog_default4 = {
   "openrouter/hunyuan-hy3": {
     providerModelName: "tencent/hy3-preview",
     aliases: [
-      "tencent/hy3-preview"
+      "tencent/hy3-preview",
+      "tencent/hy3",
+      "tencent/hy3:free"
     ],
     pricing: {
       inputPerMTok: 0.063,
@@ -20648,9 +21283,9 @@ var catalog_default4 = {
       "xiaomi/mimo-v2.5"
     ],
     pricing: {
-      inputPerMTok: 0.14,
+      inputPerMTok: 0.105,
       outputPerMTok: 0.28,
-      cacheReadPerMTok: 28e-4
+      cacheReadPerMTok: 0.028
     },
     preferredApi: "completions",
     supportedApis: [
@@ -20739,11 +21374,13 @@ var catalog_default4 = {
   "openrouter/nex-n2-pro": {
     providerModelName: "nex-agi/nex-n2-pro:free",
     aliases: [
-      "nex-agi/nex-n2-pro:free"
+      "nex-agi/nex-n2-pro:free",
+      "nex-agi/nex-n2-pro"
     ],
     pricing: {
-      inputPerMTok: 0,
-      outputPerMTok: 0
+      inputPerMTok: 0.25,
+      outputPerMTok: 1,
+      cacheReadPerMTok: 0.025
     },
     preferredApi: "completions",
     supportedApis: [
@@ -20831,7 +21468,8 @@ var catalog_default4 = {
   "openrouter/laguna-m.1": {
     providerModelName: "poolside/laguna-m.1:free",
     aliases: [
-      "poolside/laguna-m.1:free"
+      "poolside/laguna-m.1:free",
+      "poolside/laguna-m.1"
     ],
     pricing: {
       inputPerMTok: 0,
@@ -21011,7 +21649,8 @@ var catalog_default4 = {
   "openrouter/dolphin-mistral-24b-venice": {
     providerModelName: "cognitivecomputations/dolphin-mistral-24b-venice-edition:free",
     aliases: [
-      "cognitivecomputations/dolphin-mistral-24b-venice-edition:free"
+      "cognitivecomputations/dolphin-mistral-24b-venice-edition:free",
+      "cognitivecomputations/dolphin-mistral-24b-venice-edition"
     ],
     pricing: {
       inputPerMTok: 0,
@@ -21050,7 +21689,11 @@ var catalog_default4 = {
     family: "roleplay-finetunes",
     version: "1",
     status: "stable",
-    active: true
+    active: true,
+    deprecation: {
+      shutdownDate: "2026-07-19",
+      source: "openrouter"
+    }
   },
   "openrouter/mythomax-l2-13b": {
     providerModelName: "gryphe/mythomax-l2-13b",
@@ -21286,15 +21929,15 @@ var catalog_default4 = {
       "thedrummer/rocinante-12b"
     ],
     pricing: {
-      inputPerMTok: 0.17,
-      outputPerMTok: 0.43
+      inputPerMTok: 0.25,
+      outputPerMTok: 0.5
     },
     preferredApi: "completions",
     supportedApis: [
       "completions"
     ],
-    contextWindow: 32768,
-    maxOutput: 32768,
+    contextWindow: 65536,
+    maxOutput: 65536,
     capabilities: {
       toolUse: true,
       streaming: true,
@@ -21460,35 +22103,23 @@ var catalog_default4 = {
     version: "2",
     status: "legacy",
     active: true
-  }
-};
-
-// src/llm/providers/xai/catalog.json
-var catalog_default5 = {
-  "xai/grok-4.20": {
-    providerModelName: "grok-4.20-0309-reasoning",
+  },
+  "openrouter/claude-sonnet-5": {
+    providerModelName: "anthropic/claude-sonnet-5",
     aliases: [
-      "grok-4.20-0309-reasoning"
+      "anthropic/claude-sonnet-5"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.2,
-      tiers: {
-        "long-context": {
-          inputPerMTok: 2.5,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.4
-        }
-      }
+      inputPerMTok: 2,
+      outputPerMTok: 10,
+      cacheReadPerMTok: 0.2
     },
-    preferredApi: "responses",
+    preferredApi: "completions",
     supportedApis: [
-      "responses",
       "completions"
     ],
-    contextWindow: 2e6,
-    maxOutput: 2e6,
+    contextWindow: 1e6,
+    maxOutput: 128e3,
     capabilities: {
       toolUse: true,
       streaming: true,
@@ -21501,7 +22132,7 @@ var catalog_default5 = {
       videoGeneration: false
     },
     reasoning: {
-      supported: false,
+      supported: true,
       automatic: false,
       effortControl: false,
       encryptedContent: false,
@@ -21515,33 +22146,27 @@ var catalog_default5 = {
     outputModalities: [
       "text"
     ],
-    family: "grok",
-    version: "4.20",
+    family: "claude-sonnet",
+    version: "5",
     status: "stable",
     active: true
   },
-  "xai/grok-4.20-non-reasoning": {
-    providerModelName: "grok-4.20-0309-non-reasoning",
+  "openrouter/gpt-5.6-luna": {
+    providerModelName: "openai/gpt-5.6-luna",
     aliases: [
-      "grok-4.20-0309-non-reasoning"
+      "openai/gpt-5.6-luna"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.2,
-      tiers: {
-        "long-context": {
-          inputPerMTok: 2.5,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.4
-        }
-      }
+      inputPerMTok: 1,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.1
     },
-    preferredApi: "responses",
+    preferredApi: "completions",
     supportedApis: [
-      "responses",
       "completions"
     ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
     capabilities: {
       toolUse: true,
       streaming: true,
@@ -21554,7 +22179,7 @@ var catalog_default5 = {
       videoGeneration: false
     },
     reasoning: {
-      supported: false,
+      supported: true,
       automatic: false,
       effortControl: false,
       encryptedContent: false,
@@ -21568,33 +22193,332 @@ var catalog_default5 = {
     outputModalities: [
       "text"
     ],
-    family: "grok",
-    version: "4.20",
+    family: "gpt",
+    version: "5.6",
     status: "stable",
     active: true
   },
-  "xai/grok-4.20-multi-agent": {
-    providerModelName: "grok-4.20-multi-agent-0309",
+  "openrouter/gpt-5.6-terra": {
+    providerModelName: "openai/gpt-5.6-terra",
     aliases: [
-      "grok-4.20-multi-agent-0309"
+      "openai/gpt-5.6-terra"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.2,
-      tiers: {
-        "long-context": {
-          inputPerMTok: 2.5,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.4
-        }
+      inputPerMTok: 2.5,
+      outputPerMTok: 15,
+      cacheReadPerMTok: 0.25
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openrouter/gpt-5.6-sol": {
+    providerModelName: "openai/gpt-5.6-sol",
+    aliases: [
+      "openai/gpt-5.6-sol"
+    ],
+    pricing: {
+      inputPerMTok: 5,
+      outputPerMTok: 30,
+      cacheReadPerMTok: 0.5
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openrouter/gpt-5.6-luna-pro": {
+    providerModelName: "openai/gpt-5.6-luna-pro",
+    aliases: [
+      "openai/gpt-5.6-luna-pro"
+    ],
+    pricing: {
+      inputPerMTok: 1,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.1
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-pro",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openrouter/gpt-5.6-terra-pro": {
+    providerModelName: "openai/gpt-5.6-terra-pro",
+    aliases: [
+      "openai/gpt-5.6-terra-pro"
+    ],
+    pricing: {
+      inputPerMTok: 2.5,
+      outputPerMTok: 15,
+      cacheReadPerMTok: 0.25
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-pro",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openrouter/gpt-5.6-sol-pro": {
+    providerModelName: "openai/gpt-5.6-sol-pro",
+    aliases: [
+      "openai/gpt-5.6-sol-pro"
+    ],
+    pricing: {
+      inputPerMTok: 5,
+      outputPerMTok: 30,
+      cacheReadPerMTok: 0.5
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 105e4,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "gpt-pro",
+    version: "5.6",
+    status: "stable",
+    active: true
+  },
+  "openrouter/gemini-3.1-flash-lite-image": {
+    providerModelName: "google/gemini-3.1-flash-lite-image",
+    aliases: [
+      "google/gemini-3.1-flash-lite-image"
+    ],
+    pricing: {
+      inputPerMTok: 0.25,
+      outputPerMTok: 1.5
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 65536,
+    maxOutput: 66e3,
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: true,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "image",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "image"
+    ],
+    family: "gemini-image",
+    version: "3.1",
+    status: "stable",
+    active: true,
+    mediaParams: {
+      aspectRatio: {
+        values: [
+          "1:1",
+          "2:3",
+          "3:2",
+          "3:4",
+          "4:3",
+          "4:5",
+          "5:4",
+          "9:16",
+          "16:9",
+          "21:9"
+        ]
+      },
+      imageSize: {
+        values: [
+          "0.5K",
+          "1K",
+          "2K",
+          "4K"
+        ],
+        default: "1K"
       }
+    }
+  },
+  "openrouter/grok-4.5": {
+    providerModelName: "x-ai/grok-4.5",
+    aliases: [
+      "x-ai/grok-4.5"
+    ],
+    pricing: {
+      inputPerMTok: 2,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.5
     },
-    preferredApi: "responses",
+    preferredApi: "completions",
     supportedApis: [
-      "responses"
+      "completions"
     ],
-    contextWindow: 2e6,
+    contextWindow: 5e5,
     capabilities: {
       toolUse: true,
       streaming: true,
@@ -21607,7 +22531,7 @@ var catalog_default5 = {
       videoGeneration: false
     },
     reasoning: {
-      supported: false,
+      supported: true,
       automatic: false,
       effortControl: false,
       encryptedContent: false,
@@ -21622,34 +22546,1030 @@ var catalog_default5 = {
       "text"
     ],
     family: "grok",
-    version: "4.20",
+    version: "4.5",
     status: "stable",
     active: true
   },
-  "xai/grok-4.3": {
-    providerModelName: "grok-4.3",
+  "openrouter/glm-5.2": {
+    providerModelName: "z-ai/glm-5.2",
     aliases: [
-      "grok-4.3"
+      "z-ai/glm-5.2"
     ],
     pricing: {
-      inputPerMTok: 1.25,
-      outputPerMTok: 2.5,
-      cacheReadPerMTok: 0.2,
-      tiers: {
-        "long-context": {
-          inputPerMTok: 2.5,
-          outputPerMTok: 5,
-          cacheReadPerMTok: 0.4
-        }
-      }
+      inputPerMTok: 0.35,
+      outputPerMTok: 1.1,
+      cacheReadPerMTok: 0.065
     },
-    preferredApi: "responses",
+    preferredApi: "completions",
     supportedApis: [
-      "responses",
       "completions"
     ],
-    contextWindow: 1e6,
-    maxOutput: 1e6,
+    contextWindow: 1048576,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "glm",
+    version: "5.2",
+    status: "stable",
+    active: true
+  },
+  "openrouter/glm-5v-turbo": {
+    providerModelName: "z-ai/glm-5v-turbo",
+    aliases: [
+      "z-ai/glm-5v-turbo"
+    ],
+    pricing: {
+      inputPerMTok: 1.2,
+      outputPerMTok: 4,
+      cacheReadPerMTok: 0.24
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 202752,
+    maxOutput: 131072,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "vision",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "glm",
+    version: "5",
+    status: "stable",
+    active: true,
+    deprecation: {
+      shutdownDate: "2098-12-31",
+      source: "openrouter"
+    }
+  },
+  "openrouter/aion-3.0-mini": {
+    providerModelName: "aion-labs/aion-3.0-mini",
+    aliases: [
+      "aion-labs/aion-3.0-mini"
+    ],
+    pricing: {
+      inputPerMTok: 0.7,
+      outputPerMTok: 1.4,
+      cacheReadPerMTok: 0.18
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 131072,
+    maxOutput: 32768,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "aion",
+    version: "3.0",
+    status: "stable",
+    active: true
+  },
+  "openrouter/aion-3.0": {
+    providerModelName: "aion-labs/aion-3.0",
+    aliases: [
+      "aion-labs/aion-3.0"
+    ],
+    pricing: {
+      inputPerMTok: 3,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.75
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 131072,
+    maxOutput: 32768,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "aion",
+    version: "3.0",
+    status: "stable",
+    active: true
+  },
+  "openrouter/nex-n2-mini": {
+    providerModelName: "nex-agi/nex-n2-mini",
+    aliases: [
+      "nex-agi/nex-n2-mini"
+    ],
+    pricing: {
+      inputPerMTok: 0.025,
+      outputPerMTok: 0.1,
+      cacheReadPerMTok: 25e-4
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 262144,
+    maxOutput: 262144,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "nex",
+    version: "2",
+    status: "stable",
+    active: true
+  },
+  "openrouter/laguna-xs.2.1": {
+    providerModelName: "poolside/laguna-xs-2.1",
+    aliases: [
+      "poolside/laguna-xs-2.1",
+      "poolside/laguna-xs-2.1:free"
+    ],
+    pricing: {
+      inputPerMTok: 0.06,
+      outputPerMTok: 0.12,
+      cacheReadPerMTok: 0.03
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 262144,
+    maxOutput: 32768,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "code",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "laguna",
+    version: "2.1",
+    status: "preview",
+    active: true
+  },
+  "openrouter/fugu-ultra": {
+    providerModelName: "sakana/fugu-ultra",
+    aliases: [
+      "sakana/fugu-ultra"
+    ],
+    pricing: {
+      inputPerMTok: 5,
+      outputPerMTok: 30,
+      cacheReadPerMTok: 0.5
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 1e6,
+    maxOutput: 128e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false,
+      webSearch: true
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "fugu",
+    version: "1",
+    status: "stable",
+    active: true
+  },
+  "openrouter/north-mini-code": {
+    providerModelName: "cohere/north-mini-code:free",
+    aliases: [
+      "cohere/north-mini-code:free"
+    ],
+    pricing: {
+      inputPerMTok: 0,
+      outputPerMTok: 0
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 256e3,
+    maxOutput: 64e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: false,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "code",
+    inputModalities: [
+      "text"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "north-code",
+    version: "1",
+    status: "stable",
+    active: true
+  },
+  "openrouter/kimi-k2.7-code": {
+    providerModelName: "moonshotai/kimi-k2.7-code",
+    aliases: [
+      "moonshotai/kimi-k2.7-code"
+    ],
+    pricing: {
+      inputPerMTok: 0.72,
+      outputPerMTok: 3.49,
+      cacheReadPerMTok: 0.159
+    },
+    preferredApi: "completions",
+    supportedApis: [
+      "completions"
+    ],
+    contextWindow: 262144,
+    maxOutput: 262144,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: true,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "code",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "kimi-code",
+    version: "2.7",
+    status: "stable",
+    active: true
+  }
+};
+
+// src/llm/providers/xai/catalog.json
+var catalog_default5 = {
+  "xai/grok-4.20": {
+    providerModelName: "grok-4.20-0309-reasoning",
+    aliases: [
+      "grok-4.20-0309-reasoning"
+    ],
+    pricing: {
+      inputPerMTok: 1.25,
+      outputPerMTok: 2.5,
+      cacheReadPerMTok: 0.2,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 2.5,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.4
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 2e6,
+    maxOutput: 2e6,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "grok",
+    version: "4.20",
+    status: "stable",
+    active: true
+  },
+  "xai/grok-4.20-non-reasoning": {
+    providerModelName: "grok-4.20-0309-non-reasoning",
+    aliases: [
+      "grok-4.20-0309-non-reasoning"
+    ],
+    pricing: {
+      inputPerMTok: 1.25,
+      outputPerMTok: 2.5,
+      cacheReadPerMTok: 0.2,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 2.5,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.4
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "grok",
+    version: "4.20",
+    status: "stable",
+    active: true,
+    contextWindow: 2e6
+  },
+  "xai/grok-4.20-multi-agent": {
+    providerModelName: "grok-4.20-multi-agent-0309",
+    aliases: [
+      "grok-4.20-multi-agent-0309"
+    ],
+    pricing: {
+      inputPerMTok: 1.25,
+      outputPerMTok: 2.5,
+      cacheReadPerMTok: 0.2,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 2.5,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.4
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses"
+    ],
+    contextWindow: 2e6,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "grok",
+    version: "4.20",
+    status: "stable",
+    active: true
+  },
+  "xai/grok-4.3": {
+    providerModelName: "grok-4.3",
+    aliases: [
+      "grok-4.3"
+    ],
+    pricing: {
+      inputPerMTok: 1.25,
+      outputPerMTok: 2.5,
+      cacheReadPerMTok: 0.2,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 2.5,
+          outputPerMTok: 5,
+          cacheReadPerMTok: 0.4
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 1e6,
+    maxOutput: 1e6,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "chat",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "grok",
+    version: "4.3",
+    status: "stable",
+    active: true
+  },
+  "xai/grok-build-0.1": {
+    providerModelName: "grok-build-0.1",
+    aliases: [
+      "grok-build-0.1"
+    ],
+    pricing: {
+      inputPerMTok: 1,
+      outputPerMTok: 2,
+      cacheReadPerMTok: 0.2,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 2,
+          outputPerMTok: 4,
+          cacheReadPerMTok: 0.4
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 256e3,
+    capabilities: {
+      toolUse: true,
+      streaming: true,
+      structuredOutput: true,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "code",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "text"
+    ],
+    family: "grok",
+    version: "0.1",
+    status: "preview",
+    active: true
+  },
+  "xai/grok-imagine-image": {
+    providerModelName: "grok-imagine-image",
+    aliases: [
+      "grok-imagine-image"
+    ],
+    pricing: {
+      perUnit: {
+        "1k": 0.02,
+        "2k": 0.02
+      },
+      perImage: 0.02
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: true,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "image",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "image"
+    ],
+    family: "grok-imagine-image",
+    version: "1",
+    status: "stable",
+    active: true,
+    mediaParams: {
+      aspectRatio: {
+        values: [
+          "1:1",
+          "3:4",
+          "4:3",
+          "9:16",
+          "16:9",
+          "2:3",
+          "3:2",
+          "9:19.5",
+          "19.5:9",
+          "9:20",
+          "20:9",
+          "1:2",
+          "2:1",
+          "auto"
+        ]
+      },
+      resolution: {
+        values: [
+          "1k",
+          "2k"
+        ]
+      }
+    }
+  },
+  "xai/grok-imagine-image-quality": {
+    providerModelName: "grok-imagine-image-quality",
+    aliases: [
+      "grok-imagine-image-quality"
+    ],
+    pricing: {
+      perUnit: {
+        "1k": 0.05,
+        "2k": 0.07
+      },
+      perImage: 0.05
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: true,
+      audioGeneration: false,
+      videoGeneration: false
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "image",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "image"
+    ],
+    family: "grok-imagine-image",
+    version: "1",
+    status: "stable",
+    active: true,
+    mediaParams: {
+      aspectRatio: {
+        values: [
+          "1:1",
+          "3:4",
+          "4:3",
+          "9:16",
+          "16:9",
+          "2:3",
+          "3:2",
+          "9:19.5",
+          "19.5:9",
+          "9:20",
+          "20:9",
+          "1:2",
+          "2:1",
+          "auto"
+        ]
+      },
+      resolution: {
+        values: [
+          "1k",
+          "2k"
+        ]
+      }
+    }
+  },
+  "xai/grok-imagine-video": {
+    providerModelName: "grok-imagine-video",
+    aliases: [
+      "grok-imagine-video"
+    ],
+    pricing: {
+      perSecond: 0.01,
+      perUnit: {
+        "480p": 0.05,
+        "720p": 0.07
+      },
+      perImage: 2e-3
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: true
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "video",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "video"
+    ],
+    family: "grok-imagine-video",
+    version: "1",
+    status: "stable",
+    active: true,
+    mediaParams: {
+      duration: {
+        min: 1,
+        max: 15,
+        default: 8
+      },
+      aspectRatio: {
+        values: [
+          "1:1",
+          "16:9",
+          "9:16",
+          "4:3",
+          "3:4",
+          "3:2",
+          "2:3"
+        ]
+      },
+      resolution: {
+        values: [
+          "480p",
+          "720p",
+          "1080p"
+        ]
+      }
+    }
+  },
+  "xai/grok-imagine-video-1.5": {
+    providerModelName: "grok-imagine-video-1.5-preview",
+    aliases: [
+      "grok-imagine-video-1.5-preview",
+      "grok-imagine-video-1.5"
+    ],
+    pricing: {
+      perSecond: 0.01,
+      perUnit: {
+        "480p": 0.08,
+        "720p": 0.14,
+        "1080p": 0.25
+      },
+      perImage: 0.01
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    capabilities: {
+      toolUse: false,
+      streaming: true,
+      structuredOutput: false,
+      vision: true,
+      audio: false,
+      video: false,
+      imageGeneration: false,
+      audioGeneration: false,
+      videoGeneration: true
+    },
+    reasoning: {
+      supported: false,
+      automatic: false,
+      effortControl: false,
+      encryptedContent: false,
+      summaryAvailable: false
+    },
+    type: "video",
+    inputModalities: [
+      "text",
+      "image"
+    ],
+    outputModalities: [
+      "video"
+    ],
+    family: "grok-imagine-video",
+    version: "1.5",
+    status: "preview",
+    active: true,
+    mediaParams: {
+      duration: {
+        min: 1,
+        max: 15,
+        default: 8
+      },
+      aspectRatio: {
+        values: [
+          "1:1",
+          "16:9",
+          "9:16",
+          "4:3",
+          "3:4",
+          "3:2",
+          "2:3"
+        ]
+      },
+      resolution: {
+        values: [
+          "480p",
+          "720p",
+          "1080p"
+        ]
+      }
+    }
+  },
+  "xai/grok-4.5": {
+    providerModelName: "grok-4.5",
+    aliases: [
+      "grok-4.5"
+    ],
+    pricing: {
+      inputPerMTok: 2,
+      outputPerMTok: 6,
+      cacheReadPerMTok: 0.5,
+      tiers: {
+        "long-context": {
+          inputPerMTok: 4,
+          outputPerMTok: 12,
+          cacheReadPerMTok: 1
+        }
+      }
+    },
+    preferredApi: "responses",
+    supportedApis: [
+      "responses",
+      "completions"
+    ],
+    contextWindow: 5e5,
+    maxOutput: 5e5,
     capabilities: {
       toolUse: true,
       streaming: true,
@@ -21677,366 +23597,17 @@ var catalog_default5 = {
       "text"
     ],
     family: "grok",
-    version: "4.3",
+    version: "4.5",
     status: "stable",
     active: true
-  },
-  "xai/grok-build-0.1": {
-    providerModelName: "grok-build-0.1",
-    aliases: [
-      "grok-build-0.1"
-    ],
-    pricing: {
-      inputPerMTok: 1,
-      outputPerMTok: 2,
-      cacheReadPerMTok: 0.2,
-      tiers: {
-        "long-context": {
-          inputPerMTok: 2,
-          outputPerMTok: 4,
-          cacheReadPerMTok: 0.4
-        }
-      }
-    },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    contextWindow: 256e3,
-    capabilities: {
-      toolUse: true,
-      streaming: true,
-      structuredOutput: true,
-      vision: true,
-      audio: false,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: false,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "code",
-    inputModalities: [
-      "text",
-      "image"
-    ],
-    outputModalities: [
-      "text"
-    ],
-    family: "grok",
-    version: "0.1",
-    status: "preview",
-    active: true
-  },
-  "xai/grok-imagine-image": {
-    providerModelName: "grok-imagine-image",
-    aliases: [
-      "grok-imagine-image"
-    ],
-    pricing: {
-      perUnit: {
-        "1k": 2e-3,
-        "2k": 0.02
-      },
-      perImage: 0.02
-    },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: true,
-      audio: false,
-      video: false,
-      imageGeneration: true,
-      audioGeneration: false,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "image",
-    inputModalities: [
-      "text",
-      "image"
-    ],
-    outputModalities: [
-      "image"
-    ],
-    family: "grok-imagine-image",
-    version: "1",
-    status: "stable",
-    active: true,
-    mediaParams: {
-      aspectRatio: {
-        values: [
-          "1:1",
-          "3:4",
-          "4:3",
-          "9:16",
-          "16:9",
-          "2:3",
-          "3:2",
-          "9:19.5",
-          "19.5:9",
-          "9:20",
-          "20:9",
-          "1:2",
-          "2:1",
-          "auto"
-        ]
-      },
-      resolution: {
-        values: [
-          "1k",
-          "2k"
-        ]
-      }
-    }
-  },
-  "xai/grok-imagine-image-quality": {
-    providerModelName: "grok-imagine-image-quality",
-    aliases: [
-      "grok-imagine-image-quality"
-    ],
-    pricing: {
-      perUnit: {
-        "1k": 0.01,
-        "2k": 0.05
-      },
-      perImage: 0.04
-    },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: true,
-      audio: false,
-      video: false,
-      imageGeneration: true,
-      audioGeneration: false,
-      videoGeneration: false
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "image",
-    inputModalities: [
-      "text",
-      "image"
-    ],
-    outputModalities: [
-      "image"
-    ],
-    family: "grok-imagine-image",
-    version: "1",
-    status: "stable",
-    active: true,
-    mediaParams: {
-      aspectRatio: {
-        values: [
-          "1:1",
-          "3:4",
-          "4:3",
-          "9:16",
-          "16:9",
-          "2:3",
-          "3:2",
-          "9:19.5",
-          "19.5:9",
-          "9:20",
-          "20:9",
-          "1:2",
-          "2:1",
-          "auto"
-        ]
-      },
-      resolution: {
-        values: [
-          "1k",
-          "2k"
-        ]
-      }
-    }
-  },
-  "xai/grok-imagine-video": {
-    providerModelName: "grok-imagine-video",
-    aliases: [
-      "grok-imagine-video"
-    ],
-    pricing: {
-      perSecond: 0.01,
-      perUnit: {
-        "480p": 2e-3,
-        "720p": 0.05,
-        "1080p": 0.07
-      }
-    },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: true,
-      audio: false,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: false,
-      videoGeneration: true
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "video",
-    inputModalities: [
-      "text",
-      "image"
-    ],
-    outputModalities: [
-      "video"
-    ],
-    family: "grok-imagine-video",
-    version: "1",
-    status: "stable",
-    active: true,
-    mediaParams: {
-      duration: {
-        min: 1,
-        max: 15,
-        default: 8
-      },
-      aspectRatio: {
-        values: [
-          "1:1",
-          "16:9",
-          "9:16",
-          "4:3",
-          "3:4",
-          "3:2",
-          "2:3"
-        ]
-      },
-      resolution: {
-        values: [
-          "480p",
-          "720p",
-          "1080p"
-        ]
-      }
-    }
-  },
-  "xai/grok-imagine-video-1.5": {
-    providerModelName: "grok-imagine-video-1.5-preview",
-    aliases: [
-      "grok-imagine-video-1.5-preview"
-    ],
-    pricing: {
-      perSecond: 0.01,
-      perUnit: {
-        "480p": 0.08,
-        "720p": 0.14
-      }
-    },
-    preferredApi: "responses",
-    supportedApis: [
-      "responses",
-      "completions"
-    ],
-    capabilities: {
-      toolUse: false,
-      streaming: true,
-      structuredOutput: false,
-      vision: true,
-      audio: false,
-      video: false,
-      imageGeneration: false,
-      audioGeneration: false,
-      videoGeneration: true
-    },
-    reasoning: {
-      supported: false,
-      automatic: false,
-      effortControl: false,
-      encryptedContent: false,
-      summaryAvailable: false
-    },
-    type: "video",
-    inputModalities: [
-      "text",
-      "image"
-    ],
-    outputModalities: [
-      "video"
-    ],
-    family: "grok-imagine-video",
-    version: "1.5",
-    status: "preview",
-    active: true,
-    mediaParams: {
-      duration: {
-        min: 1,
-        max: 15,
-        default: 8
-      },
-      aspectRatio: {
-        values: [
-          "1:1",
-          "16:9",
-          "9:16",
-          "4:3",
-          "3:4",
-          "3:2",
-          "2:3"
-        ]
-      },
-      resolution: {
-        values: [
-          "480p",
-          "720p",
-          "1080p"
-        ]
-      }
-    }
   }
 };
 
 // src/llm/providers/builtin-tools.ts
 var PROVIDER_BUILTIN_TOOLS = {
-  anthropic: ["web_search", "code_interpreter"],
+  anthropic: ["web_search", "web_fetch", "code_interpreter"],
   openai: ["web_search", "code_interpreter"],
-  google: ["web_search", "code_interpreter"],
+  google: ["web_search", "web_fetch", "code_interpreter"],
   xai: ["web_search", "code_interpreter"],
   openrouter: ["web_search"]
 };
@@ -22591,6 +24162,20 @@ function resolveServerState(args) {
 // src/llm/client-internal.ts
 var PRIORITY_INTERACTIVE = 1;
 var PRIORITY_BACKGROUND = 2;
+function buildAssistantMessage(response, origin) {
+  const stateful = origin.api === "responses" || origin.api === "interactions";
+  return {
+    role: "assistant",
+    content: response.content,
+    id: response.id || crypto.randomUUID(),
+    createdAt: Date.now(),
+    origin: {
+      provider: origin.provider,
+      model: origin.model,
+      ...stateful && response.id ? { serverStateId: response.id } : {}
+    }
+  };
+}
 function normalizeInput(input) {
   if (typeof input === "string") {
     return [{ role: "user", content: input }];
@@ -22621,7 +24206,11 @@ function contentToText(content) {
 }
 function parseStructured(text) {
   const stripped = text.trim().replace(/^```(?:json)?\s*/i, "").replace(/```\s*$/i, "").trim();
-  return JSON.parse(stripped);
+  try {
+    return JSON.parse(stripped);
+  } catch (cause) {
+    throw new InvalidFinalOutputError(text, { cause });
+  }
 }
 function buildContext(client, options) {
   const provided = options.ctx ?? {};
@@ -22742,18 +24331,11 @@ var LLMClient = class {
    *    const r2 = await llm.complete(messages); // sends id + only the new turn
    */
   assistantMessage(response) {
-    const stateful = this.api === "responses" || this.api === "interactions";
-    return {
-      role: "assistant",
-      content: response.content,
-      id: response.id || crypto.randomUUID(),
-      createdAt: Date.now(),
-      origin: {
-        provider: this.provider,
-        model: this.model,
-        ...stateful && response.id ? { serverStateId: response.id } : {}
-      }
-    };
+    return buildAssistantMessage(response, {
+      provider: this.provider,
+      model: this.model,
+      api: this.api
+    });
   }
   // ─── Moderation helpers ───────────────────────────────────────────────
   /** Resolve the OpenAI key the emulated moderation path needs. Reuses the
@@ -22818,6 +24400,8 @@ var LLMClient = class {
       maxTokens: options.maxTokens,
       temperature: options.temperature,
       topP: options.topP,
+      presencePenalty: options.presencePenalty,
+      frequencyPenalty: options.frequencyPenalty,
       stop: options.stop,
       tools: options.tools,
       toolChoice: options.toolChoice,
@@ -22944,11 +24528,24 @@ var LLMClient = class {
    *  leading/trailing markdown fences from the model reply, then JSON.parses
    *  to T. Throws if the parse fails — callers should catch + retry. */
   async structuredComplete(input, schema, options = {}) {
-    const res = await this.complete(input, {
-      ...options,
-      structured: { ...options.structured ?? {}, schema }
-    });
-    return parseStructured(res.text);
+    const structured = { ...options.structured ?? {}, schema };
+    const repairAttempts = Math.max(0, structured.repairAttempts ?? 0);
+    const messages = normalizeInput(input);
+    for (let attempt = 0; ; attempt++) {
+      const res = await this.complete(messages, { ...options, structured });
+      try {
+        return parseStructured(res.text);
+      } catch (err) {
+        if (!(err instanceof InvalidFinalOutputError) || attempt >= repairAttempts) throw err;
+        messages.push(
+          { role: "assistant", content: res.text },
+          {
+            role: "user",
+            content: `Your previous reply was not valid JSON for the required schema (${err.message}). Reply with ONLY the JSON object matching the schema \u2014 no prose, no code fences.`
+          }
+        );
+      }
+    }
   }
   async *stream(input, options = {}) {
     if (!this.fetchStreamFn) {
@@ -22965,6 +24562,8 @@ var LLMClient = class {
       maxTokens: options.maxTokens,
       temperature: options.temperature,
       topP: options.topP,
+      presencePenalty: options.presencePenalty,
+      frequencyPenalty: options.frequencyPenalty,
       stop: options.stop,
       tools: options.tools,
       toolChoice: options.toolChoice,
@@ -23069,8 +24668,15 @@ var LLMClient = class {
         case "file":
           files.push(event.file);
           break;
-        case "builtin_tool_start":
-          builtinToolCalls.push({ tool: event.tool, ...event.id ? { id: event.id } : {} });
+        case "builtin_tool_end":
+          builtinToolCalls.push({
+            tool: event.tool,
+            ...event.id ? { id: event.id } : {},
+            ...event.code ? { code: event.code } : {},
+            ...event.output ? { output: event.output } : {},
+            ...event.query ? { query: event.query } : {},
+            ...event.url ? { url: event.url } : {}
+          });
           break;
         case "moderation":
           moderationReport = this.mergeModeration(
@@ -23336,10 +24942,12 @@ var NATIVE_TO_UNIFIED = {
   code_interpreter_call: "code_interpreter",
   // Anthropic server_tool_use names
   web_search: "web_search",
+  web_fetch: "web_fetch",
   code_execution: "code_interpreter",
   bash_code_execution: "code_interpreter",
   // Anthropic *_tool_result block types
   web_search_tool_result: "web_search",
+  web_fetch_tool_result: "web_fetch",
   code_execution_tool_result: "code_interpreter",
   bash_code_execution_tool_result: "code_interpreter"
 };
@@ -23390,6 +24998,24 @@ function filesFromCodeExecBlock(block) {
   }
   return files;
 }
+function builtinInputPayload(tool, input) {
+  if (!input) return {};
+  if (tool === "code_interpreter") {
+    const code = input.code ?? input.command;
+    return typeof code === "string" ? { code } : {};
+  }
+  if (tool === "web_search") {
+    return typeof input.query === "string" ? { query: input.query } : {};
+  }
+  if (tool === "web_fetch") {
+    return typeof input.url === "string" ? { url: input.url } : {};
+  }
+  return {};
+}
+function resultStdout(content) {
+  const c = content;
+  return c && typeof c.stdout === "string" ? c.stdout : void 0;
+}
 var AnthropicAdapter = class {
   name = "anthropic";
   apiKey;
@@ -23437,7 +25063,21 @@ var AnthropicAdapter = class {
       body.tools = req.tools.map((t, i) => {
         if (!isFunctionTool(t)) {
           if (t.type === "web_search") {
-            return { type: "web_search_20250305", name: "web_search", max_uses: 5 };
+            return {
+              type: "web_search_20260318",
+              name: "web_search",
+              max_uses: 5,
+              allowed_callers: ["direct"],
+              ...t.params
+            };
+          }
+          if (t.type === "web_fetch") {
+            return {
+              type: "web_fetch_20260318",
+              name: "web_fetch",
+              allowed_callers: ["direct"],
+              ...t.params
+            };
           }
           if (t.type === "code_interpreter") {
             return { type: "code_execution_20260521", name: "code_execution" };
@@ -23485,6 +25125,10 @@ var AnthropicAdapter = class {
     });
     const headers = {};
     if (hasFileRef) headers["anthropic-beta"] = "files-api-2025-04-14";
+    const userProfileId = req.providerOptions?.userProfileId;
+    if (typeof userProfileId === "string" && userProfileId) {
+      headers["anthropic-user-profile-id"] = userProfileId;
+    }
     const usesCodeExec = req.tools?.some(
       (t) => !isFunctionTool(t) && t.type === "code_interpreter"
     );
@@ -23572,11 +25216,20 @@ var AnthropicAdapter = class {
         content.push(tc);
         toolCalls.push(tc);
       } else if (block.type === "server_tool_use") {
+        const tool = unifiedBuiltinTool(block.name);
         builtinToolCalls.push({
-          tool: unifiedBuiltinTool(block.name),
-          ...typeof block.id === "string" ? { id: block.id } : {}
+          tool,
+          ...typeof block.id === "string" ? { id: block.id } : {},
+          ...builtinInputPayload(tool, block.input)
         });
       } else {
+        if (typeof block.type === "string" && block.type.endsWith("_tool_result")) {
+          const output = resultStdout(block.content);
+          if (output) {
+            const call = builtinToolCalls.find((c) => c.id === block.tool_use_id);
+            if (call) call.output = output;
+          }
+        }
         files.push(...filesFromCodeExecBlock(block));
       }
     }
@@ -23600,6 +25253,17 @@ var AnthropicAdapter = class {
     };
   }
   parseStreamEvent(event) {
+    return this.streamEvents(event, { pending: /* @__PURE__ */ new Map() });
+  }
+  /** Stateful — Anthropic streams `server_tool_use` input via `input_json_delta`
+   *  (empty at block start) and returns the result in a separate `*_tool_result`
+   *  block. The closure accumulates each call's input (code / query) and attaches it
+   *  to the matching `builtin_tool_end`. */
+  createStreamParser() {
+    const state = { pending: /* @__PURE__ */ new Map() };
+    return (event) => this.streamEvents(event, state);
+  }
+  streamEvents(event, state) {
     if (event.event === "ping") return [];
     const data = JSON.parse(event.data);
     const type = data.type;
@@ -23609,33 +25273,58 @@ var AnthropicAdapter = class {
       if (delta.type === "thinking_delta")
         return [{ type: "thinking", text: delta.thinking }];
       if (delta.type === "input_json_delta") {
+        if (state.current) {
+          state.current.json += delta.partial_json ?? "";
+          return [];
+        }
         return [{ type: "tool_call_delta", id: "", arguments: delta.partial_json }];
       }
     }
     if (type === "content_block_start") {
       const block = data.content_block;
       if (block.type === "tool_use") {
+        state.current = void 0;
         return [{ type: "tool_call_start", id: block.id, name: block.name }];
       }
       const events = [];
       const blockType = block.type;
       if (blockType === "server_tool_use") {
+        const tool = unifiedBuiltinTool(block.name);
+        state.current = { id: block.id ?? "", tool, json: "" };
         events.push({
           type: "builtin_tool_start",
-          tool: unifiedBuiltinTool(block.name),
+          tool,
           ...typeof block.id === "string" ? { id: block.id } : {}
         });
       } else if (blockType?.endsWith("_tool_result")) {
+        const tool = unifiedBuiltinTool(blockType);
+        const id = typeof block.tool_use_id === "string" ? block.tool_use_id : void 0;
+        const input = id ? state.pending.get(id) : void 0;
+        if (id) state.pending.delete(id);
+        const output = resultStdout(block.content);
         events.push({
           type: "builtin_tool_end",
-          tool: unifiedBuiltinTool(blockType),
-          ...typeof block.tool_use_id === "string" ? { id: block.tool_use_id } : {}
+          tool,
+          ...id ? { id } : {},
+          ...input?.code ? { code: input.code } : {},
+          ...input?.query ? { query: input.query } : {},
+          ...input?.url ? { url: input.url } : {},
+          ...output ? { output } : {}
         });
       }
       for (const file of filesFromCodeExecBlock(block)) events.push({ type: "file", file });
       return events;
     }
     if (type === "content_block_stop") {
+      if (state.current) {
+        let input = {};
+        try {
+          input = JSON.parse(state.current.json || "{}");
+        } catch {
+        }
+        state.pending.set(state.current.id, builtinInputPayload(state.current.tool, input));
+        state.current = void 0;
+      }
     }
     if (type === "message_delta") {
       const delta = data.delta;
@@ -23657,11 +25346,6 @@ var AnthropicAdapter = class {
     }
     return [];
   }
-  /** Stateless — every event is self-contained (code-execution result blocks
-   *  arrive complete in a single content_block_start). */
-  createStreamParser() {
-    return (event) => this.parseStreamEvent(event);
-  }
   parseUsage(u) {
     if (!u) return emptyUsage();
     const inputTokens = u.input_tokens ?? 0;
@@ -23972,6 +25656,8 @@ var GoogleAdapter = class {
     if (req.maxTokens) config.maxOutputTokens = req.maxTokens;
     if (req.temperature !== void 0) config.temperature = req.temperature;
     if (req.topP !== void 0) config.topP = req.topP;
+    if (req.presencePenalty !== void 0) config.presencePenalty = req.presencePenalty;
+    if (req.frequencyPenalty !== void 0) config.frequencyPenalty = req.frequencyPenalty;
     if (req.stop) config.stopSequences = req.stop;
     if (req.outputModalities?.includes("audio")) {
       config.responseModalities = ["AUDIO"];
@@ -24007,6 +25693,9 @@ var GoogleAdapter = class {
       if (req.tools.some((t) => !isFunctionTool(t) && t.type === "code_interpreter")) {
         tools.push({ codeExecution: {} });
       }
+      if (req.tools.some((t) => !isFunctionTool(t) && t.type === "web_fetch")) {
+        tools.push({ urlContext: {} });
+      }
       if (tools.length) body.tools = tools;
     }
     if (req.toolChoice) {
@@ -24170,8 +25859,32 @@ var GoogleAdapter = class {
       { MAX_TOKENS: "length", SAFETY: "content_filter" }
     );
     const builtinToolCalls = [];
-    if (hasCodeExec) builtinToolCalls.push({ tool: "code_interpreter" });
-    if (candidate.groundingMetadata) builtinToolCalls.push({ tool: "web_search" });
+    const codes = parts.filter((p) => p.executableCode?.code).map((p) => p.executableCode.code);
+    const outputs = parts.filter((p) => p.codeExecutionResult).map((p) => String(p.codeExecutionResult.output ?? ""));
+    if (codes.length) {
+      codes.forEach((code, i) => {
+        builtinToolCalls.push({
+          tool: "code_interpreter",
+          code,
+          ...outputs[i] ? { output: outputs[i] } : {}
+        });
+      });
+    } else if (hasCodeExec) {
+      builtinToolCalls.push({ tool: "code_interpreter", ...outputs[0] ? { output: outputs[0] } : {} });
+    }
+    const grounding = candidate.groundingMetadata;
+    if (grounding) {
+      const q = grounding.webSearchQueries?.[0];
+      builtinToolCalls.push({ tool: "web_search", ...typeof q === "string" ? { query: q } : {} });
+    }
+    const urlCtx = candidate.urlContextMetadata;
+    const urlMeta = urlCtx?.urlMetadata;
+    if (Array.isArray(urlMeta)) {
+      for (const m of urlMeta) {
+        const url = m.retrievedUrl;
+        builtinToolCalls.push({ tool: "web_fetch", ...typeof url === "string" ? { url } : {} });
+      }
+    }
     return {
       id: crypto.randomUUID(),
       // Google doesn't return a response ID in generateContent
@@ -24222,8 +25935,21 @@ var GoogleAdapter = class {
       if (part.text !== void 0 && !part.thought)
         events.push({ type: "text", text: part.text });
       if (part.thought && part.text) events.push({ type: "thinking", text: part.text });
-      if (part.executableCode) events.push({ type: "builtin_tool_start", tool: "code_interpreter" });
-      if (part.codeExecutionResult) events.push({ type: "builtin_tool_end", tool: "code_interpreter" });
+      if (part.executableCode) {
+        const code = part.executableCode.code;
+        state.pendingCode = typeof code === "string" ? code : void 0;
+        events.push({ type: "builtin_tool_start", tool: "code_interpreter" });
+      }
+      if (part.codeExecutionResult) {
+        const output = part.codeExecutionResult.output;
+        events.push({
+          type: "builtin_tool_end",
+          tool: "code_interpreter",
+          ...state.pendingCode ? { code: state.pendingCode } : {},
+          ...typeof output === "string" && output ? { output } : {}
+        });
+        state.pendingCode = void 0;
+      }
       if (part.inlineData) {
         const inline = part.inlineData;
         const mime = inline.mimeType;
@@ -24256,8 +25982,26 @@ var GoogleAdapter = class {
     }
     if (candidate.groundingMetadata && !state.webSearchEmitted) {
       state.webSearchEmitted = true;
+      const q = candidate.groundingMetadata.webSearchQueries?.[0];
       events.push({ type: "builtin_tool_start", tool: "web_search" });
-      events.push({ type: "builtin_tool_end", tool: "web_search" });
+      events.push({
+        type: "builtin_tool_end",
+        tool: "web_search",
+        ...typeof q === "string" ? { query: q } : {}
+      });
+    }
+    const streamUrlMeta = candidate.urlContextMetadata?.urlMetadata;
+    if (Array.isArray(streamUrlMeta) && !state.urlFetchEmitted) {
+      state.urlFetchEmitted = true;
+      for (const m of streamUrlMeta) {
+        const url = m.retrievedUrl;
+        events.push({ type: "builtin_tool_start", tool: "web_fetch" });
+        events.push({
+          type: "builtin_tool_end",
+          tool: "web_fetch",
+          ...typeof url === "string" ? { url } : {}
+        });
+      }
     }
     const fr = candidate.finishReason;
     if (fr)
@@ -24327,6 +26071,8 @@ var GoogleInteractionsAdapter = class {
     if (req.maxTokens) genConfig.max_output_tokens = req.maxTokens;
     if (req.temperature !== void 0) genConfig.temperature = req.temperature;
     if (req.topP !== void 0) genConfig.top_p = req.topP;
+    if (req.presencePenalty !== void 0) genConfig.presence_penalty = req.presencePenalty;
+    if (req.frequencyPenalty !== void 0) genConfig.frequency_penalty = req.frequencyPenalty;
     if (req.stop) genConfig.stop_sequences = req.stop;
     if (req.tools?.length) {
       body.tools = req.tools.filter(isFunctionTool).map((t) => ({
@@ -24342,6 +26088,8 @@ var GoogleInteractionsAdapter = class {
       };
     }
     if (Object.keys(genConfig).length > 0) body.generation_config = genConfig;
+    const cachedContent = req.providerOptions?.cachedContent;
+    if (typeof cachedContent === "string" && cachedContent) body.cached_content = cachedContent;
     if (req.structured) {
       body.response_format = {
         type: "text",
@@ -24499,47 +26247,77 @@ var GoogleInteractionsAdapter = class {
       raw
     };
   }
-  parseStreamEvent(event) {
+  /** Translate one Interactions SSE event to unified events. The 2.10 wire is a
+   *  step machine (verified live): `step.start` opens a typed step (`model_output`,
+   *  `function_call`, `thought`…), `step.delta` streams its payload (`{type:'text'}`,
+   *  `{type:'arguments_delta'}`, `{type:'thought_summary'}`, internal
+   *  `thought_signature`), `step.stop` closes it, and `interaction.completed` /
+   *  `interaction.failed` finish the turn (usage under `interaction.usage`). A
+   *  function call's `arguments_delta` carries no id, so we correlate it to the
+   *  currently-open call id held in `state`. */
+  streamEvents(event, state) {
     const data = JSON.parse(event.data);
     const type = data.event_type ?? data.type;
     const events = [];
-    if (type === "content.delta") {
-      const delta = data.delta;
-      if (delta?.type === "text") {
-        events.push({ type: "text", text: delta.text });
-      }
-      if (delta?.type === "function_call") {
-        events.push({
-          type: "tool_call_start",
-          id: delta.id ?? "",
-          name: delta.name ?? ""
-        });
-        if (delta.arguments) {
-          events.push({
-            type: "tool_call_delta",
-            id: delta.id ?? "",
-            arguments: JSON.stringify(delta.arguments)
-          });
+    if (type === "step.start") {
+      const step = data.step ?? {};
+      if (step.type === "function_call") {
+        const id = step.id ?? "";
+        state.callId = id;
+        state.sawToolCall = true;
+        events.push({ type: "tool_call_start", id, name: step.name ?? "" });
+        const args = step.arguments;
+        if (args && Object.keys(args).length > 0) {
+          events.push({ type: "tool_call_delta", id, arguments: JSON.stringify(args) });
         }
-        events.push({ type: "tool_call_end", id: delta.id ?? "" });
       }
+      return events;
+    }
+    if (type === "step.delta") {
+      const delta = data.delta ?? {};
+      const dtype = delta.type;
+      if (dtype === "text") {
+        events.push({ type: "text", text: delta.text ?? "" });
+      } else if (dtype === "thought_summary") {
+        events.push({ type: "thinking", text: delta.text ?? "" });
+      } else if (dtype === "arguments_delta") {
+        events.push({ type: "tool_call_delta", id: state.callId ?? "", arguments: delta.arguments ?? "" });
+      }
+      return events;
+    }
+    if (type === "step.stop") {
+      if (state.callId) {
+        events.push({ type: "tool_call_end", id: state.callId });
+        state.callId = void 0;
+      }
+      return events;
     }
-    if (type === "interaction.complete") {
+    if (type === "interaction.completed" || type === "interaction.failed") {
+      if (state.callId) {
+        events.push({ type: "tool_call_end", id: state.callId });
+        state.callId = void 0;
+      }
       const interaction = data.interaction ?? {};
-      const usage = interaction.usage;
+      const usage = interaction.usage ?? data.metadata?.total_usage;
       if (usage) events.push({ type: "usage", usage: this.parseUsage(usage) });
       events.push({
         type: "done",
-        finishReason: extractFinishReason(false, interaction.status, {
+        finishReason: extractFinishReason(state.sawToolCall, interaction.status, {
           failed: "error"
         })
       });
+      return events;
     }
     return events;
   }
-  /** Stateless — the Interactions adapter surfaces no code-execution file outputs. */
+  parseStreamEvent(event) {
+    return this.streamEvents(event, { sawToolCall: false });
+  }
+  /** Per-stream stateful parser — correlates a function call's streamed arguments
+   *  and end to the call opened by its `step.start`. */
   createStreamParser() {
-    return (event) => this.parseStreamEvent(event);
+    const state = { sawToolCall: false };
+    return (event) => this.streamEvents(event, state);
   }
   parseUsage(u) {
     if (!u) return emptyUsage();
@@ -25270,6 +27048,8 @@ var OpenAIAdapter = class {
     };
     if (req.temperature !== void 0) body.temperature = req.temperature;
     if (req.topP !== void 0) body.top_p = req.topP;
+    if (req.presencePenalty !== void 0) body.presence_penalty = req.presencePenalty;
+    if (req.frequencyPenalty !== void 0) body.frequency_penalty = req.frequencyPenalty;
     if (req.stop) body.stop = req.stop;
     const tier = openaiRequestTier(req.serviceTier);
     if (tier) body.service_tier = tier;
@@ -26025,32 +27805,98 @@ function filenameForMime(mimeType) {
   if (mimeType.startsWith("image/")) return `file.${mimeType.slice("image/".length)}`;
   return "file.bin";
 }
+function codeOutputFromResponsesItem(item) {
+  const parts = [];
+  for (const out of item.outputs ?? []) {
+    if (out.type !== "logs" || typeof out.logs !== "string") continue;
+    try {
+      const j = JSON.parse(out.logs);
+      if (typeof j.stdout === "string") {
+        parts.push(j.stdout);
+        continue;
+      }
+    } catch {
+    }
+    parts.push(out.logs);
+  }
+  return parts.length ? parts.join("") : void 0;
+}
+function searchActionPayload(item) {
+  const action = item.action;
+  if (!action) return {};
+  const out = {};
+  if (Array.isArray(action.queries) && typeof action.queries[0] === "string")
+    out.query = action.queries[0];
+  else if (typeof action.query === "string") out.query = action.query;
+  if (typeof action.url === "string") out.url = action.url;
+  return out;
+}
 var RESPONSES_BUILTIN_ITEMS = /* @__PURE__ */ new Set(["web_search_call", "code_interpreter_call"]);
 function builtinCallFromResponsesItem(item) {
   const type = item.type;
   if (!RESPONSES_BUILTIN_ITEMS.has(type)) return null;
+  const call = { tool: unifiedBuiltinTool(type) };
+  if (typeof item.id === "string") call.id = item.id;
+  if (type === "code_interpreter_call") {
+    if (typeof item.code === "string") call.code = item.code;
+    const output = codeOutputFromResponsesItem(item);
+    if (output) call.output = output;
+  } else if (type === "web_search_call") {
+    const { query, url } = searchActionPayload(item);
+    if (query) call.query = query;
+    if (url) call.url = url;
+  }
+  return call;
+}
+function builtinEndPayload(call) {
   return {
-    tool: unifiedBuiltinTool(type),
-    ...typeof item.id === "string" ? { id: item.id } : {}
+    ...call.id ? { id: call.id } : {},
+    ...call.code ? { code: call.code } : {},
+    ...call.output ? { output: call.output } : {},
+    ...call.query ? { query: call.query } : {},
+    ...call.url ? { url: call.url } : {}
   };
 }
+var IMAGE_EXTS = /* @__PURE__ */ new Set(["png", "jpg", "jpeg", "gif", "webp", "svg", "bmp"]);
+function fileExt(name) {
+  const i = name.lastIndexOf(".");
+  return i >= 0 ? name.slice(i + 1).toLowerCase() : "";
+}
+function isDisplayArtifact(c) {
+  if (!c.filename) return false;
+  const ext = fileExt(c.filename);
+  return c.filename === `${c.fileId}.${ext}` && IMAGE_EXTS.has(ext) && c.span[0] === c.span[1];
+}
 function filesFromResponsesOutputItem(item) {
   const files = [];
   const type = item.type;
   if (type === "message") {
+    const citations = [];
     for (const c of item.content ?? []) {
       if (c.type !== "output_text") continue;
       for (const a of c.annotations ?? []) {
         if (a.type === "container_file_citation" && typeof a.file_id === "string") {
-          files.push({
-            id: a.file_id,
-            ...typeof a.filename === "string" ? { name: a.filename } : {},
-            ...typeof a.container_id === "string" ? { ref: { containerId: a.container_id } } : {},
-            source: "code_execution"
+          citations.push({
+            fileId: a.file_id,
+            filename: typeof a.filename === "string" ? a.filename : void 0,
+            containerId: typeof a.container_id === "string" ? a.container_id : void 0,
+            span: [Number(a.start_index) || 0, Number(a.end_index) || 0]
           });
         }
       }
     }
+    const hasSavedImage = citations.some(
+      (c) => !isDisplayArtifact(c) && IMAGE_EXTS.has(fileExt(c.filename ?? ""))
+    );
+    for (const c of citations) {
+      if (hasSavedImage && isDisplayArtifact(c)) continue;
+      files.push({
+        id: c.fileId,
+        ...c.filename ? { name: c.filename } : {},
+        ...c.containerId ? { ref: { containerId: c.containerId } } : {},
+        source: "code_execution"
+      });
+    }
   }
   if (type === "code_interpreter_call") {
     for (const out of item.outputs ?? []) {
@@ -26354,7 +28200,7 @@ var OpenAIResponsesAdapter = class {
       for (const file of this.filesFromOutputItem(item)) events.push({ type: "file", file });
       const builtin = builtinCallFromResponsesItem(item ?? {});
       if (builtin) {
-        events.push({ type: "builtin_tool_end", tool: builtin.tool, ...builtin.id ? { id: builtin.id } : {} });
+        events.push({ type: "builtin_tool_end", tool: builtin.tool, ...builtinEndPayload(builtin) });
       }
       if (item?.type === "function_call") {
         events.push({ type: "tool_call_end", id: item.call_id ?? "" });
@@ -26992,6 +28838,14 @@ var XAIMediaAdapter = class {
   }
 };
 
+// src/llm/providers/xai/tiers.ts
+function xaiRequestTier(t) {
+  if (!t) return void 0;
+  if (t === "priority") return "priority";
+  if (t === "standard" || t === "default") return "default";
+  return void 0;
+}
+
 // src/llm/providers/xai/responses.ts
 function xaiCodeExecFiles(item) {
   if (item.type !== "code_interpreter_call") return [];
@@ -27035,6 +28889,9 @@ var XAIResponsesAdapter = class extends OpenAIResponsesAdapter {
     if (!req.model.includes("multi-agent")) {
       delete body.reasoning;
     }
+    const xaiTier = xaiRequestTier(req.serviceTier);
+    if (xaiTier) body.service_tier = xaiTier;
+    else delete body.service_tier;
     const usesCodeInterpreter = req.tools?.some(
       (t) => !isFunctionTool(t) && t.type === "code_interpreter"
     );
@@ -28253,6 +30110,7 @@ var AgentLoop = class _AgentLoop {
     let lastResponse = null;
     let reason = "done";
     let errorMsg;
+    let caughtError;
     let guardrailTripReason;
     try {
       while (true) {
@@ -28301,7 +30159,11 @@ var AgentLoop = class _AgentLoop {
         totalLlmTimeMs += stepLatency;
         addUsage(totalUsage, lastResponse.usage);
         this._history.append(
-          { role: "assistant", content: lastResponse.content },
+          buildAssistantMessage(lastResponse, {
+            provider: this.client.provider,
+            model: this.client.model,
+            api: this.client.api
+          }),
           { model: this.client.model, usage: lastResponse.usage, latencyMs: stepLatency }
         );
         const hasToolCalls = lastResponse.finishReason === "tool_use" && lastResponse.toolCalls.length > 0;
@@ -28355,6 +30217,7 @@ var AgentLoop = class _AgentLoop {
     } catch (e) {
       reason = "error";
       errorMsg = e instanceof Error ? e.message : String(e);
+      caughtError = e;
       await this.hooks.emit("onRunError", {
         runId,
         agentId: this.id,
@@ -28405,6 +30268,7 @@ var AgentLoop = class _AgentLoop {
       totalToolTimeMs,
       runTrace
     });
+    if (reason === "error") throw caughtError ?? new Error(errorMsg ?? "agent run failed");
     return finalResponse;
   }
   /** Run `complete` with a JSON Schema enforced via `structured`, then
@@ -28431,6 +30295,7 @@ var AgentLoop = class _AgentLoop {
     let lastResponse = null;
     let reason = "done";
     let errorMsg;
+    let caughtError;
     let guardrailTripReason;
     try {
       while (true) {
@@ -28492,7 +30357,14 @@ var AgentLoop = class _AgentLoop {
         finalText = state.stepText;
         lastResponse = stepResponse;
         this._history.append(
-          { role: "assistant", content },
+          {
+            ...buildAssistantMessage(lastResponse, {
+              provider: this.client.provider,
+              model: this.client.model,
+              api: this.client.api
+            }),
+            content
+          },
           { model: this.client.model, usage: state.stepUsage, latencyMs: stepLatency }
         );
         const hasToolCalls = state.stepToolCalls.length > 0;
@@ -28554,6 +30426,7 @@ var AgentLoop = class _AgentLoop {
     } catch (e) {
       reason = "error";
       errorMsg = e instanceof Error ? e.message : String(e);
+      caughtError = e;
       await this.hooks.emit("onRunError", {
         runId,
         agentId: this.id,
@@ -28602,6 +30475,7 @@ var AgentLoop = class _AgentLoop {
       totalToolTimeMs,
       runTrace
     });
+    if (reason === "error") throw caughtError ?? new Error(errorMsg ?? "agent run failed");
     yield { type: "done", response: finalResponse };
   }
   /** Yield tool_call_start events for all tool calls in this step. */
@@ -36983,6 +38857,7 @@ var CAP_KEYS = {
 var BUILTIN_TOOL_KEYS = {
   search: "web_search",
   web_search: "web_search",
+  web_fetch: "web_fetch",
   code_interpreter: "code_interpreter"
 };
 var KNOWN_KEYS = /* @__PURE__ */ new Set([
@@ -38532,6 +40407,7 @@ function createRetrieval(backend, config) {
 export {
   AgentBus,
   AgentLoop,
+  AgentRunError,
   AnthropicAdapter,
   AnthropicBatchAdapter,
   AnthropicCountApi,
@@ -38587,6 +40463,7 @@ export {
   INPUT_SIZE_BUCKET_LABELS,
   InMemoryVectorStore,
   InternalToolRunner,
+  InvalidFinalOutputError,
   JSON_API_SYSTEM_PROMPT,
   LAYER_AGENTLOOP_CONTEXT,
   LAYER_AGENTLOOP_SYSTEM,