Atlas

Benchmarks

← All benchmarks

TxBench-PP

Science · 2026-06-17

Preclinical biology benchmark spanning disease models, screening, target validation, pharmacodynamics, efficacy, and safety tasks.

Top models (higher is better)

ModelScore
Opus 4.859.3
GPT-5.555.3
Gemini 3.5 Flash51.3
GPT-5.449.7
Opus 4.749.3
Opus 4.644.7
Gemini 3.1 Pro Preview40.0
Sonnet 4.636.0
Kimi K2.629.7
Grok 4.2019.7
Grok 4.318.3
Loading Atlas data…