metadata
title: REST API endpoints for models inference
shortTitle: Inference
intro: >-
Use the REST API to submit a chat completion request to a specified model,
with or without organizational attribution.
versions:
fpt: '*'
topics:
- API
autogenerated: rest
allowTitleToDifferFromFilename: true
About {% data variables.product.prodname_github_models %} inference
You can use the REST API to run inference requests using the {% data variables.product.prodname_github_models %} platform. The API requires the models: read scope when using a {% data variables.product.pat_v2 %} or when authenticating using a {% data variables.product.prodname_github_app %}.
The API supports:
- Accessing top models from OpenAI, DeepSeek, Microsoft, Llama, and more.
- Running chat-based inference requests with full control over sampling and response parameters.
- Streaming or non-streaming completions.
- Organizational attribution and usage tracking.