Calibration
BrowseComp → AA AutomationBench
AA AutomationBench is estimated from BrowseComp with a offset logistic curve fitted on 6 models measured on both: y = 0.0256 + (0.7955 − 0.0256) / (1 + exp(−31.84·(x − 0.8712))), R² = 0.99, cross-validated error 5.7 pp. It is used for 29 estimates.
| Estimated model | BrowseComp | AA AutomationBench | Source |
|---|---|---|---|
| Agents-A1 | 75.5% | 4.4% | estimated ± 5.7 pp, low confidence |
| Agents-A1-4B | 66.8% | 2.7% | estimated ± 5.7 pp, low confidence |
| Claude Mythos 5 | 88.0% | 46.4% | estimated ± 5.7 pp, low confidence |
| Claude Opus 4.8 | 84.3% | 24.8% | estimated ± 5.7 pp, low confidence |
| Claude Opus 5 | 90.8% | 61.3% | estimated ± 5.7 pp, low confidence |
| Claude Sonnet 5 | 84.7% | 26.9% | estimated ± 5.7 pp, low confidence |
| DeepSeek V4 Flash 0731 | 73.2% | 3.5% | estimated ± 5.7 pp, low confidence |
| DeepSeek V4 Pro 0813 | 83.4% | 20.6% | estimated ± 5.7 pp, low confidence |
| dots3-note Preview | 83.3% | 20.2% | estimated ± 5.7 pp, low confidence |
| GLM-4.7 | 52.0% | 2.6% | estimated ± 5.7 pp, low confidence |
| GLM-5.1 | 68.0% | 2.7% | estimated ± 5.7 pp, low confidence |
| GPT-5.4 Pro | 89.3% | 53.9% | estimated ± 5.7 pp, low confidence |
| GPT-5.5 Pro | 90.1% | 58.0% | estimated ± 5.7 pp, low confidence |
| Inkling-Small | 77.4% | 5.9% | estimated ± 5.7 pp, low confidence |
| Kimi K2.6 | 83.2% | 19.7% | estimated ± 5.7 pp, low confidence |
| Kimi K2.5 (Reasoning) | 60.6% | 2.6% | estimated ± 5.7 pp, low confidence |
| Ling 3.0 Flash | 72.2% | 3.2% | estimated ± 5.7 pp, low confidence |
| LongCat-Flash-Lite-Sparse | 48.6% | 2.6% | estimated ± 5.7 pp, low confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 36.8% | 2.6% | estimated ± 5.7 pp, low confidence |
| Ornith-1.5-35B-A3B | 67.6% | 2.7% | estimated ± 5.7 pp, low confidence |
| Ornith-1.5-397B | 86.6% | 37.9% | estimated ± 5.7 pp, low confidence |
| Ornith-1.5-9B | 56.4% | 2.6% | estimated ± 5.7 pp, low confidence |
| Qwen3.5-122B-A10B | 63.8% | 2.6% | estimated ± 5.7 pp, low confidence |
| Qwen3.5-27B | 61.0% | 2.6% | estimated ± 5.7 pp, low confidence |
| Qwen3.5-35B-A3B | 61.0% | 2.6% | estimated ± 5.7 pp, low confidence |
| Qwen3.5 397B | 62.0% | 2.6% | estimated ± 5.7 pp, low confidence |
| Beam | 77.4% | 5.9% | estimated ± 5.7 pp, low confidence |
| Solar Pro 4 | 49.2% | 2.6% | estimated ± 5.7 pp, low confidence |
| Step 3.7 Flash | 75.8% | 4.6% | estimated ± 5.7 pp, low confidence |