Test fixes in openai_compat

docs/_static/llama-stack-spec.yaml

@@ -4548,6 +4548,8 @@ components:
             $ref: '#/components/schemas/TokenLogProbs'
           description: >-
             Optional log probabilities for generated tokens
+        usage:
+          $ref: '#/components/schemas/UsageInfo'
       additionalProperties: false
       required:
         - completion_message
@@ -4589,6 +4591,25 @@ components:
         - logprobs_by_token
       title: TokenLogProbs
       description: Log probabilities for generated tokens.
+    UsageInfo:
+      type: object
+      properties:
+        completion_tokens:
+          type: integer
+          description: Number of tokens generated
+        prompt_tokens:
+          type: integer
+          description: Number of tokens in the prompt
+        total_tokens:
+          type: integer
+          description: Total number of tokens processed
+      additionalProperties: false
+      required:
+        - completion_tokens
+        - prompt_tokens
+        - total_tokens
+      title: UsageInfo
+      description: Usage information for a model.
     BatchCompletionRequest:
       type: object
       properties:
@@ -8103,6 +8124,26 @@ components:
       title: OpenAIChatCompletionToolCallFunction
       description: >-
         Function call details for OpenAI-compatible tool calls.
+    OpenAIChatCompletionUsage:
+      type: object
+      properties:
+        prompt_tokens:
+          type: integer
+          description: The number of tokens in the prompt
+        completion_tokens:
+          type: integer
+          description: The number of tokens in the completion
+        total_tokens:
+          type: integer
+          description: The total number of tokens used
+      additionalProperties: false
+      required:
+        - prompt_tokens
+        - completion_tokens
+        - total_tokens
+      title: OpenAIChatCompletionUsage
+      description: >-
+        Usage information for an OpenAI-compatible chat completion response.
     OpenAIChoice:
       type: object
       properties:
@@ -8365,6 +8406,12 @@ components:
     OpenAICompletionWithInputMessages:
       type: object
       properties:
+        metrics:
+          type: array
+          items:
+            $ref: '#/components/schemas/MetricInResponse'
+          description: >-
+            (Optional) List of metrics associated with the API response
         id:
           type: string
           description: The ID of the chat completion
@@ -8387,6 +8434,8 @@ components:
           type: string
           description: >-
             The model that was used to generate the chat completion
+        usage:
+          $ref: '#/components/schemas/OpenAIChatCompletionUsage'
         input_messages:
           type: array
           items:
@@ -9682,6 +9731,12 @@ components:
           items:
             type: object
             properties:
+              metrics:
+                type: array
+                items:
+                  $ref: '#/components/schemas/MetricInResponse'
+                description: >-
+                  (Optional) List of metrics associated with the API response
               id:
                 type: string
                 description: The ID of the chat completion
@@ -9704,6 +9759,8 @@ components:
                 type: string
                 description: >-
                   The model that was used to generate the chat completion
+              usage:
+                $ref: '#/components/schemas/OpenAIChatCompletionUsage'
               input_messages:
                 type: array
                 items:
@@ -10719,6 +10776,12 @@ components:
     OpenAIChatCompletion:
       type: object
       properties:
+        metrics:
+          type: array
+          items:
+            $ref: '#/components/schemas/MetricInResponse'
+          description: >-
+            (Optional) List of metrics associated with the API response
         id:
           type: string
           description: The ID of the chat completion
@@ -10741,6 +10804,8 @@ components:
           type: string
           description: >-
             The model that was used to generate the chat completion
+        usage:
+          $ref: '#/components/schemas/OpenAIChatCompletionUsage'
       additionalProperties: false
       required:
         - id