# OpenAI's MedChemBench finds models rebuild molecular structures reliably but score about 50% on compound selection

The internal benchmark draws 99 problems from 117 papers and patents, and OpenAI warns that those sources may be in the models' training data.

- Published: 2026-11-21T06:18:37.672Z
- Canonical: https://polylog.news/ai/2026-11-21/openai-s-medchembench-finds-models-rebuild-molecular-structu
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [OpenAI News](https://openai.com/index/introducing-medchembench/), [AI Weekly](https://aiweekly.co/alerts/openai-releases-medchembench-gpt-62-astra-leads-at-845), [Superpower Daily](https://superpowerdaily.com/posts/openai-releases-chemistry-benchmark-showing-ai-s-uneven-drug-research-skills)

OpenAI introduced MedChemBench on November 20. The benchmark tests whether models can interpret molecular structures, reason from experimental evidence and select compounds in early drug discovery. It contains 99 problems drawn from 117 pap…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-11-21/openai-s-medchembench-finds-models-rebuild-molecular-structu (subscription information: https://polylog.news/pricing).