Dynamic Task Success on DynaCU-Bench
82Success RateClaude Sonnet 4.6
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Claude Sonnet 4.6Observation Interface=AOI (full), Model Source=Closed-source2026.06 | 82 | 73.3 | 44 | 1.3 | |
| Gemini 2.5 FlashObservation Interface=AOI (full), Model Source=Closed-source2026.06 | 69 | 59.4 | 48 | 2.9 | |
| Grok-4.3Observation Interface=AOI (full), Model Source=Closed-source2026.06 | 65 | 55.3 | 40 | 8.2 | |
| GPT-5.4Observation Interface=AOI (full), Model Source=Closed-source2026.06 | 57 | 47.2 | 20 | 8.8 | |
| EvoCUA-32BObservation Interface=AOI (full), Model Source=Open-source2026.06 | 55 | 45.2 | 37 | 3 | |
| Grok-4-fast-reasoningObservation Interface=AOI (full), Model Source=Closed-source2026.06 | 47 | 37.5 | 28 | 4.9 | |
| Gemini 3 FlashObservation Interface=AOI (full), Model Source=Closed-source2026.06 | 45 | 35.6 | 9 | 0.18 | |
| Claude Sonnet 4.6Observation Interface=Standard, Model Source=Closed-source2026.06 | 38 | 29.1 | — | — | |
| GPT-5.4Observation Interface=Standard, Model Source=Closed-source2026.06 | 37 | 28.2 | — | — | |
| Gemini 3 FlashObservation Interface=Standard, Model Source=Closed-source2026.06 | 36 | 27.3 | — | — | |
| Fara-7BObservation Interface=AOI (full), Model Source=Open-source2026.06 | 34 | 25.5 | 17 | 4.9 | |
| Grok-4Observation Interface=AOI (full), Model Source=Closed-source2026.06 | 25 | 17.5 | 21 | 3 | |
| Grok-4.3Observation Interface=Standard, Model Source=Closed-source2026.06 | 25 | 17.5 | — | — | |
| Gemini 2.5 FlashObservation Interface=Standard, Model Source=Closed-source2026.06 | 21 | 14.2 | — | — | |
| Grok-4-fast-reasoningObservation Interface=Standard, Model Source=Closed-source2026.06 | 19 | 12.5 | — | — | |
| EvoCUA-32BObservation Interface=Standard, Model Source=Open-source2026.06 | 18 | 11.7 | — | — | |
| Fara-7BObservation Interface=Standard, Model Source=Open-source2026.06 | 17 | 10.9 | — | — | |
| Grok-4Observation Interface=Standard, Model Source=Closed-source2026.06 | 4 | 1.6 | — | — |