TxBench Oligo Discovery V2
Science
TxBench-Oligo-Discovery V2 clean-v1 reruns on 120 experimental-data evaluations. Macro-mean pass rate; the aggregate site's weight of 113 is not the task count.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 49.2 |
| Claude Opus 5 | 45.6 |
| Grok 4.6 | 43.3 |
| GPT-6 Sol | 40.6 |
| Grok 4.7 | 40.3 |
| Gemini 3.7 Flash | 35.6 |
| Opus 4.8 | 35.3 |
| GPT-5.6 Sol | 35.0 |
| Sonnet 5 | 34.4 |
| Gemini 3.8 Flash | 33.1 |
| GPT-5.5 | 31.9 |
| GPT-6 Luna | 31.7 |
| Opus 4.7 | 30.0 |
| Sonnet 4.6 | 28.3 |
| GPT-5.6 Luna | 21.7 |
| Claude Opus 5.5 | 1.4 |