Atlas

Benchmarks

← All benchmarks

TxBench Oligo Discovery V1 (113 evaluations)

Science · 2026-10-09

The October 9, 2026 published V1 oligo-discovery cohort contains 113 evaluation records, reduced from the previously published 120-record cohort. Macro-mean pass rates are kept separate from that earlier cohort.

Top models (higher is better)

ModelScore
GPT-6.1 Sol52.5
Claude Sonnet 5.551.9
GPT-6 Astra51.9
Claude Opus 548.4
Grok 4.646.0
GPT-6 Sol43.1
Grok 4.741.9
Gemini 3.7 Flash37.8
Opus 4.837.5
GPT-5.6 Sol37.2
Sonnet 536.6
Gemini 3.8 Flash35.1
DeepSeek V4.1 Flash34.8
GPT-5.533.9
GPT-6 Luna33.6
Opus 4.731.9
Kimi K331.3
Sonnet 4.630.1
GPT-5.6 Luna23.0
Nemotron 3.5 Lightning 30B A3B (precision unspecified)8.6
Claude Opus 5.51.5
Loading Atlas data…