chore: update OpenAPI spec from monorepo (#640)

Co-authored-by: OpenRouter SDK Bot <sdk-bot@openrouter.ai>

Author: openrouter-docs-sync[bot]

.speakeasy/in.openapi.yaml

@@ -9073,8 +9073,9 @@ components:
         - 'title'
       type: 'object'
     GenerationContentData:
-      description: 'Stored prompt and completion content'
+      description: 'Stored prompt and completion content, plus the failure error when one was stored'
       example:
+        error: null
         input:
           messages:
             - content: 'What is the meaning of life?'
@@ -9083,6 +9084,8 @@ components:
           completion: 'The meaning of life is a philosophical question...'
           reasoning: null
       properties:
+        error:
+          $ref: '#/components/schemas/GenerationContentError'
         input:
           anyOf:
             - properties:
@@ -9125,11 +9128,98 @@ components:
       required:
         - 'input'
         - 'output'
+        - 'error'
+      type: 'object'
+    GenerationContentError:
+      description: 'The stored failure for this generation, or null when it succeeded'
+      example:
+        message: 'Timed out waiting for the provider'
+        previous_errors:
+          - code: 429
+            message: 'Provider returned error'
+            provider_name: 'Google'
+            raw: '{"error":{"code":429,"message":"Resource exhausted"}}'
+        provider_name: 'Vertex'
+        raw: '{"error":{"code":504,"message":"Deadline exceeded"}}'
+        status: 504
+      properties:
+        message:
+          description: 'Error message returned to the client'
+          example: 'Timed out waiting for the provider'
+          type:
+            - 'string'
+            - 'null'
+        previous_errors:
+          description: 'Every upstream attempt that failed before the returned error, in attempt order'
+          items:
+            $ref: '#/components/schemas/GenerationContentErrorAttempt'
+          type: 'array'
+        provider_name:
+          description: 'Provider whose error was returned to the client, when known'
+          example: 'Vertex'
+          type:
+            - 'string'
+            - 'null'
+        raw:
+          description: 'Raw error body behind the returned error, when stored'
+          example: '{"error":{"code":504,"message":"Deadline exceeded"}}'
+          type:
+            - 'string'
+            - 'null'
+        status:
+          description: 'HTTP status returned to the client'
+          example: 504
+          type:
+            - 'integer'
+            - 'null'
+      required:
+        - 'status'
+        - 'message'
+        - 'provider_name'
+        - 'raw'
+        - 'previous_errors'
+      type:
+        - 'object'
+        - 'null'
+    GenerationContentErrorAttempt:
+      description: 'A single failed upstream attempt recorded for the generation'
+      example:
+        code: 429
+        message: 'Provider returned error'
+        provider_name: 'Google'
+        raw: '{"error":{"code":429,"message":"Resource exhausted"}}'
+      properties:
+        code:
+          description: 'HTTP status returned by this attempt'
+          example: 429
+          type: 'integer'
+        message:
+          description: 'Error message returned by this attempt'
+          example: 'Provider returned error'
+          type: 'string'
+        provider_name:
+          description: 'Provider that served this attempt, when known'
+          example: 'Google'
+          type:
+            - 'string'
+            - 'null'
+        raw:
+          description: 'Raw error body returned by the provider, when stored'
+          example: '{"error":{"code":429,"message":"Resource exhausted"}}'
+          type:
+            - 'string'
+            - 'null'
+      required:
+        - 'code'
+        - 'message'
+        - 'provider_name'
+        - 'raw'
       type: 'object'
     GenerationContentResponse:
       description: 'Stored prompt and completion content for a generation'
       example:
         data:
+          error: null
           input:
             messages:
               - content: 'What is the meaning of life?'
@@ -30110,16 +30200,26 @@ paths:
             application/json:
               example:
                 data:
+                  error:
+                    message: 'Timed out waiting for the provider'
+                    previous_errors:
+                      - code: 429
+                        message: 'Provider returned error'
+                        provider_name: 'Google'
+                        raw: '{"error":{"code":429,"message":"Resource exhausted"}}'
+                    provider_name: 'Vertex'
+                    raw: '{"error":{"code":504,"message":"Deadline exceeded"}}'
+                    status: 504
                   input:
                     messages:
                       - content: 'What is the meaning of life?'
                         role: 'user'
                   output:
-                    completion: 'The meaning of life is a philosophical question...'
+                    completion: null
                     reasoning: null
               schema:
                 $ref: '#/components/schemas/GenerationContentResponse'
-          description: 'Returns the stored prompt and completion content'
+          description: 'Returns the stored prompt and completion content, plus the failure error when the generation failed'
         '401':
           content:
             application/json:
@@ -30200,7 +30300,7 @@ paths:
               schema:
                 $ref: '#/components/schemas/ProviderOverloadedResponse'
           description: 'Provider Overloaded - Provider is temporarily overloaded'
-      summary: 'Get stored prompt and completion content for a generation'
+      summary: 'Get stored prompt, completion, and error content for a generation'
       tags:
         - 'Generations'
   /generation/feedback: