Benchmarks
357 sources · 1,263 models × 3,079 benchmarks · 162,127 scoresOpenAI'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. Answers reconstruct molecules as SMILES (761 distinct graded structures), explain results, rank or select compounds and predict measured properties. Deterministic checks (RDKit-canonical isomeric SMILES matches, pairwise ranking and Jaccard set credit, fold-error credit for numbers) and a GPT-6 Astra judge at medium effort for rationale criteria give partial credit, and the score is the share of each task's rubric budget earned, averaged over ten responses per task (completed invalid responses score zero) and then equally over the 99 tasks. OpenAI built it and reports only its own GPT models. It is not the 'MedChemBench (Internal)' of OpenAI's launch posts (medchembench-openai-internal-reported), whose values differ.
57th strongest of 818 fitted benchmarks
Signal-to-noise +16.94
278th hardest of 818 fitted benchmarks
Capability needed 159.6
| # | Model | Setting | Score | Source |
|---|---|---|---|---|
| 1 | GPT-6.2 Astra | x-high | 84.5 | |
| 2 | GPT-6 Astra | max | 83.8 | |
| 3 | GPT-6.1 Sol | max | 82.9 | |
| 4 | GPT-6.1 Terra | max | 77.6 | |
| 5 | GPT-6 Sol | max | 69.1 | |
| 6 | GPT-5.6 Sol | max | 59.8 | |
| 7 | GPT-5.5 | x-high | 49.3 | |
| 8 | GPT-5.6 Terra | max | 43.3 | |
| 9 | GPT-5.6 Luna | max | 40.6 | |
| 10 | GPT-6 Luna | max | 40.1 | |
| 11 | GPT-5.4 | x-high | 39.2 | |
| 12 | GPT-5.2 | x-high | 37.8 |
Benchmarks outside its family whose model rankings track it most reliably (Spearman ρ over at least 10 shared models).