currently, does it support openai or deepseek api?

[q1andq2]

I can see prompt format are not json.

[mssssss123]

We use the standard OpenAI interface, so it should support various API-compatible models, including OpenAI and DeepSeek.
The prompt templates we provide are based on the Jina format, but this does not affect using OpenAI or DeepSeek APIs. 😊

[q1andq2]

In generation.py line 125, It got Error code while calling openai because of the following , which is not standard OpenAI parameters as it comments. should a vLLM-specific configuration for that?

125: # Separate vLLM-specific parameters from standard OpenAI parameters
openai_params = {k: v for k, v in sampling_params.items() if k != 'extra_body'}
extra_body = sampling_params.get('extra_body', {})

                resp = await client.chat.completions.create(
                    model=model_name,
                    messages=msg,
                    extra_body=extra_body,
                    **openai_params,
                )

[q1andq2]

Got it, there is unrelated parameters in yaml