Runtime capability checks for provider features that cannot be inferred safely.
Token logprob support is determined from the response metadata. The current ReqLLM integration supports OpenAI Chat Completions responses with a nonempty logprob list. OpenAI Responses, Anthropic, Google/Gemini, and responses without logprob metadata return an explicit unsupported reason.
Summary
Functions
Checks whether metadata proves token logprob support for this response.
Types
Functions
@spec token_logprobs(map()) :: :ok | {:error, unavailable_reason()}
Checks whether metadata proves token logprob support for this response.