ResearchMedium importance
OpenAI Launches MedChemBench with 99 Medicinal Chemistry Problems; GPT-6.2 Astra Earns 84.5% of Credit
OpenAI presented MedChemBench this Friday (11/20), a benchmark with 99 medicinal chemistry problems that measures whether models link molecular structures to experimental evidence to explain results, compare compounds, and predict properties, according to Unite.AI. The tasks cover pre-clinical drug discovery. GPT-6.2 Astra achieved 84.5% of the available credit at the xhigh reasoning level. GPT-6 Astra secured 95.37% of the molecular structure credit, but only 50.42% in candidate ranking and selection. OpenAI itself admits that the benchmark is nearing saturation and does not show that models improve real-world drug discovery decisions.
Published in edition no. 42 of Noticiário do Joãozinho ·
Translated by Joãozinho AI from the original Brazilian Portuguese edition. Facts, numbers and sources are the same as in the original; quotes, video titles and source names stay in their original language. Read the original in Portuguese
Original source
https://www.unite.ai/openai-introduces-medchembench-for-medicinal-chemistry-reasoning/