fix: change logprobs type from bool to int in Completions endpoint (#5343)

Fixes #5253

The `logprobs` field in `OpenAICompletionRequestWithExtraBody` was typed
as `bool` but the OpenAI Completions API expects an integer (0–5)
indicating how many top log probabilities to include per output token.
This caused a 400 error when callers passed an int:

```
openai.BadRequestError: ... 'Input should be a valid boolean, unable to interpret input'
```

Changes:
- `models.py`: `logprobs: bool | None` → `logprobs: int | None` with
`ge=0, le=5`
- `llama-stack-spec.yaml`: matching schema update (`boolean` → `integer`
with min/max)

The Chat Completions model (`OpenAIChatCompletionRequestWithExtraBody`)
is unchanged — its `logprobs` is correctly a boolean there.

---------

Signed-off-by: Extra Small <littleshuai.bot@gmail.com>
Signed-off-by: Sébastien Han <seb@redhat.com>
Co-authored-by: Sébastien Han <seb@redhat.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: 3 people

client-sdks/stainless/openapi.yml

@@ -5008,9 +5008,11 @@ components:
           description: The logit bias to use.
         logprobs:
           anyOf:
-          - type: boolean
+          - type: integer
+            maximum: 5.0
+            minimum: 0.0
           - type: 'null'
-          description: The log probabilities to use.
+          description: Include the log probabilities on the logprobs most likely output tokens.
         max_tokens:
           anyOf:
           - type: integer

docs/static/deprecated-ogx-spec.yaml

@@ -1311,9 +1311,11 @@ components:
           description: The logit bias to use.
         logprobs:
           anyOf:
-          - type: boolean
+          - type: integer
+            maximum: 5.0
+            minimum: 0.0
           - type: 'null'
-          description: The log probabilities to use.
+          description: Include the log probabilities on the logprobs most likely output tokens.
         max_tokens:
           anyOf:
           - type: integer

docs/static/experimental-ogx-spec.yaml

@@ -1752,9 +1752,11 @@ components:
           description: The logit bias to use.
         logprobs:
           anyOf:
-          - type: boolean
+          - type: integer
+            maximum: 5.0
+            minimum: 0.0
           - type: 'null'
-          description: The log probabilities to use.
+          description: Include the log probabilities on the logprobs most likely output tokens.
         max_tokens:
           anyOf:
           - type: integer

docs/static/ogx-spec.yaml

@@ -4565,9 +4565,11 @@ components:
           description: The logit bias to use.
         logprobs:
           anyOf:
-          - type: boolean
+          - type: integer
+            maximum: 5.0
+            minimum: 0.0
           - type: 'null'
-          description: The log probabilities to use.
+          description: Include the log probabilities on the logprobs most likely output tokens.
         max_tokens:
           anyOf:
           - type: integer

docs/static/stainless-ogx-spec.yaml

@@ -5008,9 +5008,11 @@ components:
           description: The logit bias to use.
         logprobs:
           anyOf:
-          - type: boolean
+          - type: integer
+            maximum: 5.0
+            minimum: 0.0
           - type: 'null'
-          description: The log probabilities to use.
+          description: Include the log probabilities on the logprobs most likely output tokens.
         max_tokens:
           anyOf:
           - type: integer

src/ogx_api/inference/models.py

@@ -948,7 +948,13 @@ class OpenAICompletionRequestWithExtraBody(BaseModel, extra="allow"):
         default=None, ge=-2.0, le=2.0, description="The penalty for repeated tokens."
     )
     logit_bias: dict[str, float] | None = Field(default=None, description="The logit bias to use.")
-    logprobs: bool | None = Field(default=None, description="The log probabilities to use.")
+    logprobs: int | None = Field(
+        default=None,
+        ge=0,
+        le=5,
+        strict=True,
+        description="Include the log probabilities on the logprobs most likely output tokens.",
+    )
     max_tokens: int | None = Field(default=None, ge=1, description="The maximum number of tokens to generate.")
     n: int | None = Field(default=None, ge=1, description="The number of completions to generate.")
     presence_penalty: float | None = Field(