ScienceAgentBench
Free while signed in. Answers cite the passages they came from.
First page
The curator’s take
a new benchmark to rigorously assess agents built for scientific workflows; after testing it on open-weight and proprietary LLMs, the best-performing agent can only solve 32.4% of the tasks independently and 34.3% with expert-provided knowledge.
Every Monday
Get next week’s papers.
The same picks and the same summaries, in your inbox. Free, and 176 issues deep.
Subscribe on Substack