Atlas

Benchmarks

MedChemBench (scientific rationale)

MedChemBench's scientific rationale component (Table S2 of the paper): on the 91 of 99 tasks with scientific rationale criteria, the share of rationale points earned, each criterion scored zero, half or full credit by a GPT-6 Astra judge at medium effort against task-specific anchors, averaged over ten responses per task and then equally over those tasks. It is normalized within the component rather than being an additive share of the overall score, so it is a diagnostic of medchembench. OpenAI's MedChemBench (November 20, 2026) poses 99 medicinal-chemistry problems built from 117 recent papers and patents (median publication year 2025), each giving the biological context, compound structures and experimental evidence (structure drawings, assay tables, plots, protein–ligand views) up to a decision point in a drug program; a model answers each in one turn, without tools.

Models
11
Top score
91.7
Leader
GPT-6 Astra
Updated
Nov 20, 2026
Resolution
—
Saturation
—

No signal or difficulty estimate. Held out of the fit by design.

Organizations
11 dated models · 4 frontier modelsClosed weightsOpen weights
Data: OpenAIatlas.village.dev

MedChemBench (scientific rationale) leaderboard

Related benchmarks

Loading Atlas data…