Task 3 on BlenderBench
8.98PLClaude-Sonnet-4
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Claude-Sonnet-4Setting=VIGA2026.01 | 8.98 | 11.97 | 1.8 | 53.07 | |
| Qwen3-VL-8BSetting=VIGA2026.01 | 11.35 | 6.86 | 2.25 | 112.79 | |
| GPT-4oSetting=VIGA2026.01 | 14.51 | 17.91 | 1.53 | 113.96 | |
| Qwen3-VL-8BSetting=One-Shot2026.01 | 22.81 | 22.98 | 1.25 | — | |
| Claude-Sonnet-4Setting=One-Shot2026.01 | 26.66 | 32.8 | 1.44 | — | |
| Gemini-2.5-ProSetting=VIGA2026.01 | 27.41 | 30.97 | 0.66 | 41.12 | |
| GPT-4oSetting=One-Shot2026.01 | 30.14 | 38.69 | 0.25 | — | |
| Gemini-2.5-ProSetting=One-Shot2026.01 | 31.66 | 39.89 | 0.64 | — |