SUMMARY IN gpt-5-mini-2025-08-07

API Bugs
1

Hello, I’m here to report a bug I just found using the gpt-5-mini-2025-08-07 model.

Caused by: com.stel.assistant.infrastructure.exception.AssistantException: {
“error”: {
“message”: “Invalid value: ‘null’. Supported values are: ‘concise’, ‘detailed’, and ‘auto’.”,
“type”: “invalid_request_error”,
“param”: “reasoning.summary”,
“code”: “invalid_value”
}
}

but when i use “concise “ with the same model :

“error”: {
“message”: “Unsupported value: ‘concise’ is not supported with the ‘gpt-5-mini-2025-08-07’ model. Supported values are: ‘detailed’.”,
“type”: “invalid_request_error”,
“param”: “reasoning.summary”,
“code”: “unsupported_value”
}
}

2

It’s only really a bug in the quality of message returned.

There’s several rounds of validation: the API, then the model’s accepted parameter, or sometimes reversed to catch models vs endpoints.

  • The first is telling you not to send null to the endpoint. Note that null is not a string in JSON.

summary - string or null - Optional
A summary of the reasoning performed by the model. This can be useful for debugging and understanding the model’s reasoning process. One of auto, concise, or detailed.

I ran against the API Python SDK:

reasoning = {
    "effort": None,  # gives default
    "summary": None,  # gives no events
}

No issues. If really pressed, I could code up a RESTful JSON call, with the requisite stream event handler and what-have-you.

  • The second is telling you not to send ‘concise’ to gpt-5. They don’t document clearly why the confusing choices, but “concise” is just for computer-use-preview or codex, methinks.