OpenAI releases MedChemBench, a medicinal chemistry benchmark: 99 problems from 117 papers and patents, with GPT-6.2 Astra earning 84.5% of available credit
It tests whether models can connect molecular structure with experimental evidence to decide which compound to test next. How it is graded and how model generations compare.
// Key points
- 99 problems drawn from 117 papers and patents, covering 761 distinct molecular structures; it grades molecular structures, scientific reasoning, ranking and selection, and property predictions.
- At the xhigh reasoning setting, GPT-6.2 Astra earned 84.5%, GPT-6 Astra 83.3%, GPT-6.1 Sol 82.0% and GPT-5.6 Sol 58.2%.
- On one task that requires preserving an 18-atom ring, GPT-6 Astra was correct in all ten responses, while GPT-5.6 Sol and GPT-5.6 Luna failed in all ten.
Builder's takeFor most product teams the point isn't drug discovery but how to build a vertical benchmark: drawn from real decisions, partial credit, structural errors scored on their own. The 6.x generations sit within about two and a half points of each other, so upgrades now buy little on tasks like this. If you build a vertical product, a small benchmark in this shape is worth making.
// Background · from #OpenAI
Full timeline →- Nov 21 OpenAI details its defenses against adversarial distillation: attacks recovering GPT-6.2 Astra's protected reasoning succeed 0.2% of the time, down from 54.0%
- Nov 21 OpenAI describes its Performance Factory: a Codex-driven regression workflow tracked over 1,700 investigations and root-caused over 600 in just over three months
- Nov 18 OpenAI releases GPT-6.2 Astra and GPT-6.2 Sol, with better professional outputs, coding, computer use and efficiency at lower cost