CL-Bench Cohort Studies Gain
Agents · 2026-05-04
Mean cumulative gain over the stateless baseline across 20 sequential instances of the CL-Bench cohort studies task.
Top models (higher is better)
| Model | Score |
|---|---|
| Sonnet 4.6 | 0.8 |
| GPT-5.4 | 0.4 |
| Gemini 3 Flash Preview | 0.4 |
| Opus 4.7 | -0.1 |
| Gemini 3.1 Pro Preview | -0.8 |