Calibration
AutomationBench → EnterpriseOps-Gym
EnterpriseOps-Gym is estimated from AutomationBench with a Michaelis–Menten curve fitted on 11 models measured on both: y = 0.4321·x / (0.01355 + x), R² = 0.37, cross-validated error 6.2 pp. It is used for 11 estimates.
| Estimated model | AutomationBench | EnterpriseOps-Gym | Source |
|---|---|---|---|
| GPT-6 Astra (high) | 66.6% | 42.3% | estimated ± 6.2 pp, low confidence |
| GPT-6 Astra (xhigh) | 67.2% | 42.4% | estimated ± 6.2 pp, low confidence |
| Gemini 4 Argon (high) | 77.5% | 42.5% | estimated ± 6.2 pp, low confidence |
| GPT-6 Sol (max) | 61.6% | 42.3% | estimated ± 6.2 pp, low confidence |
| MiMo-V2.6-Pro | 58.6% | 42.2% | estimated ± 6.2 pp, low confidence |
| DeepSeek V4.1 Flash (max) | 68.9% | 42.4% | estimated ± 6.2 pp, low confidence |
| Mistral Large 4 Preview | 59.9% | 42.3% | estimated ± 6.2 pp, low confidence |
| Grok 4.7 (xhigh) | 65.6% | 42.3% | estimated ± 6.2 pp, low confidence |
| GPT-6 Luna (max) | 53.2% | 42.1% | estimated ± 6.2 pp, low confidence |
| GPT-6.1 Sol (xhigh) | 66.6% | 42.3% | estimated ± 6.2 pp, low confidence |
| Grok 4.6 (xhigh) | 67.0% | 42.4% | estimated ± 6.2 pp, low confidence |