MedChemBench - OpenAI internal release score
Science
Internal medicinal-chemistry release evaluation, with undisclosed task cohort and aggregation. Diagnostic; benchmark-name similarity is not evidence of equivalence with a public corpus.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 49.3 |
| GPT-5.6 Sol | 48.3 |
| GPT-5.6 Terra | 35.0 |
| GPT-5.6 Luna | 30.4 |