OpenAI MedChemBench - Numerical Prediction: leaderboard
OpenAI's MedChemBench (20 November 2026): 99 problems built from 117 medicinal chemistry papers and patents, where a model gets a drug program's target, goals, compounds and results so far and must reconstruct structures, explain findings, pick compounds and predict properties. One attempt per response, no tools, ten responses per problem. This board is the prediction component only: how close predicted values and categories are to those reported in the source, with partial credit by closeness. Scored on 21 of the 99 problems; no model reaches 48%. Rows are model and reasoning-effort pairs for eleven OpenAI models; GPT-5.2, GPT-5.4 and GPT-5.5 come from the paper's tables. Share of rubric credit (%); higher is better.
Metric: Average credit (%). Source: openai.com. Saturation forecast: Around 2030. 11 models tracked. Updated 21 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | GPT-6 Astra | Max | 47.57 | #10 |
| 2 | GPT-6.1 Sol | Max | 46.76 | #15 |
| 3 | GPT-5.5 | High | 45.96 | #54 |
| 4 | GPT-6.1 Terra | Max | 45.55 | #25 |
| 5 | GPT-6 Sol | Max | 45.08 | #34 |
| 6 | GPT-5.6 Sol | Max | 44.46 | #37 |
| 7 | GPT-5.6 Terra | Max | 41.75 | #71 |
| 8 | GPT-5.4 | xHigh | 40.77 | #135 |
| 9 | GPT-6 Luna | Max | 39.99 | #105 |
| 10 | GPT-5.2 | xHigh | 39.99 | #185 |
| 11 | GPT-5.6 Luna | Max | 38.59 | #131 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=openai-medchembench-numerical-prediction · How It Works · Data refreshed daily, snapshot 2026-11-21.