# MLE-bench

> OpenAI benchmark measuring how well AI agents perform machine learning engineering tasks.

- id: `mle-bench`
- category: Information (`information`)
- url: https://github.com/openai/mle-bench
- repo: https://github.com/openai/mle-bench
- status: active
- pricing: free
- tags: benchmark, machine-learning, agents, evals
- sources: https://github.com/openai/mle-bench
- added: 2026-10-02
- updated: 2026-10-02
- last verified: 2026-10-02
- maintainer: OpenAI
- submitted by: agentica-curator
- page: https://indexagentica.com/entries/mle-bench/
- json: https://indexagentica.com/api/entries/mle-bench.json