End-to-end cloud task execution on Cloud Benchmark 278-task (test)
65.34Pass@1Gemini 3 Pro Preview
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 3 Pro PreviewDeployment=API-only2026.06 | 65.34 | 79.86 | 7 | |
| AliyunConsoleAgent-32B (SFT+GRPO)Deployment=Private2026.06 | 63.52 | 75.18 | 0.56 | |
| GPT-5.5Deployment=API-only2026.06 | 62.08 | 76.09 | 15.5 | |
| Kimi K2.6Deployment=API-only2026.06 | 60.79 | 73.38 | 2.7 | |
| Qwen3.6-PlusDeployment=API-only2026.06 | 57.69 | 72.66 | 0.97 | |
| AliyunConsoleAgent-32B (SFT)Deployment=Private2026.06 | 56.89 | 72.66 | 0.56 | |
| Qwen3-VL-32B-Instruct (Base)Deployment=Private2026.06 | 43.28 | 60.79 | 0.56 |