openai

/

gpt-oss-120b

like 3.5k

Follow

OpenAI 17.9k

Hello,

When calling the Hugging Face Inference API for openai/gpt-oss-120b, I’m getting this error:

"Bad Request: The endpoint is paused, ask a maintainer to restart it"
Could you please restart the endpoint?

I’m also specifically testing with huggingface_hub.InferenceClient.text_generation using:

details=True
decoder_input_details=True

to retrieve per-token log probabilities and, if possible, top_tokens for each generation step.

Could you please confirm:

Whether logprobs or top_tokens are supported for this model via the Inference API.

If yes, whether they work for both standard completions and chat-style prompts.

Any special configuration needed on my side to enable this functionality.