Corebench Easy
Fresh
CORE-Bench is a benchmark for evaluating the ability of agents to computationally reproduce scientific papers.
- Type
- RL Env
- Runtime
ORS- License
- unknown
- Size
- 18 tasks
- Published
- Feb 2026
- Canonical
- openreward.ai/siegelz/corebench-easy
Cite
Notes
Only stored in your browser.