RE-Bench

Evals

Benchmark / dataset

7 day-long ML research-engineering tasks vs human experts.

Category
Autonomy, R&D & self-replication
Insurance exposure
Unmapped
Publisher
METR
Publisher type
Nonprofit / research org
License
MIT
Status
Stale (>12mo)
Last commit
2025-10-16
Stars
171

View source on GitHub →

Wanich Risk Lab, 2026.