---
title: ML Intern API
emoji: 🛰️
colorFrom: yellow
colorTo: gray
sdk: docker
app_port: 7860
pinned: false
license: apache-2.0
short_description: OpenAI-compatible API for the ML Intern agent
---

# ML Intern API

OpenAI Responses-API-compatible HTTP API for [ML Intern](https://huggingface.co/spaces/smolagents/ml-intern):
send a task with your Hugging Face token as a Bearer header, the agent
researches/codes/launches HF Jobs under **your** namespace, and streams
results + artifacts (trackio dashboards, jobs, models) over resumable SSE.

Interactive API documentation is served separately at
https://huggingface-ml-intern-api-docs.static.hf.space so docs-only changes do not
restart this backend Space. Endpoints live under `/v1` — start at
`POST /v1/responses`.

The docs include verified public-endpoint examples, starting with short
research/model-selection tasks and extending to longer artifact-producing runs.

Source: https://github.com/huggingface/ml-intern (see `docs/API.md`).

Optional Space secrets:
- `MONGODB_URI` (+ `MONGODB_DB`) — durable event replay, reconnects, and
  restart recovery. Without it, response tracking is in-memory only.
- `HF_TOKEN` — server-side fallbacks such as session trace uploads.