Currently, the OpenAI-compatible API uses a default temperature=1.0 when generating responses. This leads to inconsistent results—especially for use cases requiring deterministic output (e.g., code generation). Users have no way to lower the temperature for more stable responses.
Could you add support for an optional temperature parameter in API requests, or set the default temperature to 0？
Currently, the OpenAI-compatible API uses a default temperature=1.0 when generating responses. This leads to inconsistent results—especially for use cases requiring deterministic output (e.g., code generation). Users have no way to lower the temperature for more stable responses.
Could you add support for an optional temperature parameter in API requests, or set the default temperature to 0？