CL-Bench Blind Spectrum Monitoring Gain
Agents · 2026-05-04
Mean cumulative gain over the stateless baseline across 90 sequential instances of the CL-Bench blind spectrum monitoring task.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.4 | 26.4 |
| Sonnet 4.6 | 24.5 |
| Opus 4.7 | 13.8 |
| Gemini 3 Flash Preview | 13.3 |
| Gemini 3.1 Pro Preview | 13.3 |