We’ve been using GPT-4o with Function Calling and Structured Outputs for our conversational AI.
We’d like to switch to gpt-5-chat (without reasoning), but at the moment these features don’t seem to be supported in the API.
Are there any plans or timelines for adding Function Calling and Structured Outputs to gpt-5-chat?
chatgpt-4o-latest - this model will point to the latest GPT-4o model used by ChatGPT.
Then gpt-5-chat-latest is released without fanfare.
The lack of features is intentional:
| feature | ChatGPT-4o | GPT-5-chat | GPT-5 |
|---|---|---|---|
| v1/chat/completions | Yes | Yes | Yes |
| v1/responses | Yes | Yes | Yes |
| v1/assistants | No | No | Yes |
| v1/batch | No | No | Yes |
| Streaming | Yes | Yes | Yes |
| Function calling | No | No | Yes |
| Structured outputs | No | No | Yes |
| Fine-tuning | No | No | Yes |
| Distillation | No | No | Yes |
| Predicted outputs | Yes | Yes | Yes |
| Image input | Yes | Yes | Yes |
The OpenAI table for gpt-5 features is wrong, considering that gpt-5 cannot be used with assistants, is not a model to be submitted to any fine-tuning method, and predicted outputs is also not allowed:
HTTP error 400: {
“error”: {
“message”: “Unsupported parameter: ‘prediction’ is not supported with this model.”,
“type”: “invalid_request_error”,
“param”: “prediction”,
“code”: “unsupported_parameter”
}
}
They quickly killed off json_schema, but you can still use json_object for “json mode” with gpt-5-chat for unenforced structured outputs.
Plus, the distillation idea was abandoned.
