Hugging Face (gateway)

Route Hugging Face model calls through the Respan gateway after the Hugging Face model you want is available in your Respan model list. Your application uses the OpenAI-compatible Respan endpoint and only sends RESPAN_API_KEY; provider credentials are managed in Respan.

Setup

Install packages

pip install openai

Set environment variables

export RESPAN_API_KEY="YOUR_RESPAN_API_KEY"

Configure the provider credentials required by your model in Respan. Once they are stored, no HUGGINGFACE_API_KEY is needed in application code. Check the exact model’s billing label in the model catalog; hosting a model on Hugging Face does not by itself make it available through Respan credits.

Point an OpenAI-compatible client to the Respan gateway

import os
from openai import OpenAI
client = OpenAI(
    api_key=os.environ["RESPAN_API_KEY"],
    base_url="https://api.respan.ai/api",
)
response = client.chat.completions.create(
    model="YOUR_HUGGINGFACE_MODEL_ID",
    messages=[{"role": "user", "content": "Say hello in three languages."}],
)
print(response.choices[0].message.content)

View your request

Open the Spans page to inspect the gateway request, model, latency, token usage, cost, and response content.

Switch models

Change the model parameter to use 1000+ models from different providers through the same gateway.

response = client.chat.completions.create(model="YOUR_HUGGINGFACE_MODEL_ID", messages=messages)
response = client.chat.completions.create(model="openai/gpt-6-sol", messages=messages)
response = client.chat.completions.create(model="anthropic/claude-opus-5-5", messages=messages)

See the full model list.