feat: support for GPT-5-Pro highest reasoning model https://github.com/BeehiveInnovations/zen-mcp-server/issues/275

providers/openai_compatible.py

@@ -391,9 +391,10 @@ class OpenAICompatibleProvider(ModelProvider):
         messages: list,
         temperature: float,
         max_output_tokens: Optional[int] = None,
+        capabilities: Optional[ModelCapabilities] = None,
         **kwargs,
     ) -> ModelResponse:
-        """Generate content using the /v1/responses endpoint for o3-pro via OpenAI library."""
+        """Generate content using the /v1/responses endpoint for reasoning models."""
         # Convert messages to the correct format for responses endpoint
         input_messages = []
 
@@ -412,10 +413,14 @@ class OpenAICompatibleProvider(ModelProvider):
 
         # Prepare completion parameters for responses endpoint
         # Based on OpenAI documentation, use nested reasoning object for responses endpoint
+        effort = "medium"
+        if capabilities and capabilities.default_reasoning_effort:
+            effort = capabilities.default_reasoning_effort
+
         completion_params = {
             "model": model_name,
             "input": input_messages,
-            "reasoning": {"effort": "medium"},  # Use nested object for responses endpoint
+            "reasoning": {"effort": effort},
             "store": True,
         }
 
@@ -475,11 +480,11 @@ class OpenAICompatibleProvider(ModelProvider):
                 operation=_attempt,
                 max_attempts=max_retries,
                 delays=retry_delays,
-                log_prefix="o3-pro responses endpoint",
+                log_prefix="responses endpoint",
             )
         except Exception as exc:
             attempts = max(attempt_counter["value"], 1)
-            error_msg = f"o3-pro responses endpoint error after {attempts} attempt{'s' if attempts > 1 else ''}: {exc}"
+            error_msg = f"responses endpoint error after {attempts} attempt{'s' if attempts > 1 else ''}: {exc}"
             logging.error(error_msg)
             raise RuntimeError(error_msg) from exc
 
@@ -614,6 +619,7 @@ class OpenAICompatibleProvider(ModelProvider):
                 messages=messages,
                 temperature=temperature,
                 max_output_tokens=max_output_tokens,
+                capabilities=capabilities,
                 **kwargs,
             )
 

providers/openai_provider.py

@@ -31,7 +31,7 @@ class OpenAIModelProvider(OpenAICompatibleProvider):
             max_output_tokens=128_000,  # 128K max output tokens
             supports_extended_thinking=True,  # Supports reasoning tokens
             supports_system_prompts=True,
-            supports_streaming=True,
+            supports_streaming=False,
             supports_function_calling=True,
             supports_json_mode=True,
             supports_images=True,  # GPT-5 supports vision
@@ -41,6 +41,27 @@ class OpenAIModelProvider(OpenAICompatibleProvider):
             description="GPT-5 (400K context, 128K output) - Advanced model with reasoning support",
             aliases=["gpt5"],
         ),
+        "gpt-5-pro": ModelCapabilities(
+            provider=ProviderType.OPENAI,
+            model_name="gpt-5-pro",
+            friendly_name="OpenAI (GPT-5 Pro)",
+            intelligence_score=18,
+            use_openai_response_api=True,
+            context_window=400_000,
+            max_output_tokens=272_000,
+            supports_extended_thinking=True,
+            supports_system_prompts=True,
+            supports_streaming=False,
+            supports_function_calling=True,
+            supports_json_mode=True,
+            supports_images=True,
+            max_image_size_mb=20.0,
+            supports_temperature=True,
+            temperature_constraint=TemperatureConstraint.create("fixed"),
+            default_reasoning_effort="high",
+            description="GPT-5 Pro (400K context, 272K output) - Advanced model with reasoning support",
+            aliases=["gpt5pro"],
+        ),
         "gpt-5-mini": ModelCapabilities(
             provider=ProviderType.OPENAI,
             model_name="gpt-5-mini",
@@ -50,7 +71,7 @@ class OpenAIModelProvider(OpenAICompatibleProvider):
             max_output_tokens=128_000,  # 128K max output tokens
             supports_extended_thinking=True,  # Supports reasoning tokens
             supports_system_prompts=True,
-            supports_streaming=True,
+            supports_streaming=False,
             supports_function_calling=True,
             supports_json_mode=True,
             supports_images=True,  # GPT-5-mini supports vision
@@ -284,7 +305,7 @@ class OpenAIModelProvider(OpenAICompatibleProvider):
         if category == ToolModelCategory.EXTENDED_REASONING:
             # Prefer models with extended thinking support
             # GPT-5-Codex first for coding tasks
-            preferred = find_first(["gpt-5-codex", "o3", "o3-pro", "gpt-5"])
+            preferred = find_first(["gpt-5-codex", "gpt-5-pro", "o3", "o3-pro", "gpt-5"])
             return preferred if preferred else allowed_models[0]
 
         elif category == ToolModelCategory.FAST_RESPONSE:
@@ -296,5 +317,5 @@ class OpenAIModelProvider(OpenAICompatibleProvider):
         else:  # BALANCED or default
             # Prefer balanced performance/cost models
             # Include GPT-5-Codex for coding workflows
-            preferred = find_first(["gpt-5", "gpt-5-codex", "gpt-5-mini", "o4-mini", "o3-mini"])
+            preferred = find_first(["gpt-5", "gpt-5-codex", "gpt-5-pro", "gpt-5-mini", "o4-mini", "o3-mini"])
             return preferred if preferred else allowed_models[0]

providers/shared/model_capabilities.py

@@ -51,6 +51,7 @@ class ModelCapabilities:
     supports_json_mode: bool = False
     supports_temperature: bool = True
     use_openai_response_api: bool = False
+    default_reasoning_effort: Optional[str] = None
 
     # Additional attributes
     max_image_size_mb: float = 0.0