NetGoodIndexSubmit a correction

Benefit Ledger · verified · Chemistry

ChemCrow uses GPT-4 plus tools to execute organic-chemistry tasks

EPFL and collaborators published ChemCrow in Nature Machine Intelligence on 8 May 2024, an agent that used GPT-4 and 18 chemistry tools to plan syntheses including an insect repellent and organocatalysts.

8 May 2024Tier 2 NotableMethodology 0.1

Current score

+0.16

3 base · Notable (tier 2 of 5, 3 pts)
× 0.5500 attribution · Critical contribution
× 0.7000 evidence · Peer review or independent validation
× 0.5000 realization · Experimentally validated
× 0.4500 durability
Event-level product before credit split: 0.26

Notable chemistry-agent demonstration (tier 2). Shared credit with tools and humans. Peer-reviewed. Lab-scale realization. Quickly superseded agent stacks.

What happened

Expert and LLM evaluations found the tool-using agent more reliable than an unaided LLM on chemistry tasks. Demonstrations included autonomous planning and, in some cases, execution guidance. It is a research prototype, not a replacement for process chemistry.

Model attribution

+0.16

GPT

Reasoning LLM that called ChemCrow tools.

GPT-4 was the controller; 18 expert tools and human evaluation were essential.

Attribution 0.5500 · Credit share 60% · OpenAI

Claims

  • ChemCrow planned and in reported cases executed syntheses of small molecules using GPT-4 plus chemistry tools.

    outcome · supported

Sources

primary sources

Secondary domains: Computer Science

Revision history

  • 13 Sep 2026 · 0.00 → 0.16

    Initial adjudicated seed score under methodology 0.1.