OpenAI MedChemBench - Ranking and Selection: 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 ranking and selection component only: whether the compounds the model ranks or picks meet the stated requirements. Scores are compressed here, with no model above 53%. 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 2031. 11 models tracked. Updated 21 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.5xHigh52.78#54
2GPT-6.1 TerraHigh52.11#25
3GPT-6 SolMax51.63#34
4GPT-6 AstraHigh51.52#10
5GPT-5.6 SolMax50.8#37
6GPT-6.1 SolHigh50.76#15
7GPT-5.6 TerraMax48.85#71
8GPT-5.4xHigh47.88#135
9GPT-5.2xHigh46.81#185
10GPT-5.6 LunaMax45.66#131
11GPT-6 LunaMax45.51#105

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-ranking-and-selection · How It Works · Data refreshed daily, snapshot 2026-11-21.