Single-step operation description prediction on Android-In-The-Zoo
82.1AATeach VLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Teach VLM2026.06 | 82.1 | 67.66 | 92.91 | 58.84 | |
| Doubao-Seed-2.02026.06 | 64.08 | 37.04 | 42.37 | 42.37 | |
| Doubao-Seed-1.82026.06 | 61.66 | 35.37 | 56.85 | 30.28 | |
| Qwen3-VL-8B2026.06 | 41.35 | 27.19 | 57.22 | 23.07 | |
| Qwen3.5-9B2026.06 | 22.71 | 25.55 | 29.81 | 17.38 | |
| GPT-4o2026.06 | 20.93 | 23.6 | 12.7 | 16.07 |