Video Object Segmentation on Colon-Bench 272 videos (test)
48.3mIoUGemini 3 Flash
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini 3 Flashvariant=Flash2026.03 | 48.3 | 54.7 | |
| Gemini 3 Provariant=Pro2026.03 | 45 | 51.3 | |
| Gemini 3.1 Flash Litevariant=Flash Lite2026.03 | 37.4 | 43.4 | |
| GPT-5.42026.03 | 34.5 | 41.1 | |
| GPT-5.22026.03 | 30.7 | 36.5 | |
| Qwen-VL Maxvariant=Max2026.03 | 25.6 | 29.6 | |
| Qwen3-VL Plusvariant=Plus2026.03 | 20.4 | 24.6 | |
| Gemini 2.5 Flash Litevariant=Flash Lite2026.03 | 19.9 | 24.3 | |
| Qwen3.5 Plusvariant=Plus2026.03 | 16.7 | 21 | |
| Qwen3.5 397Bparameters=397B2026.03 | 16.6 | 21 | |
| Claude Opus 4.6variant=Opus2026.03 | 16.1 | 20.5 | |
| Qwen3-VL 235Bparameters=235B2026.03 | 13.6 | 16.9 | |
| Qwen3-VL 32Bparameters=32B2026.03 | 12.7 | 15.9 | |
| Seed 1.62026.03 | 12.6 | 16.1 | |
| GLM-4.6V2026.03 | 12.5 | 16.1 | |
| Qwen3-VL 8Bparameters=8B2026.03 | 10.4 | 13.1 | |
| Seed 1.6 Flashvariant=Flash2026.03 | 2.6 | 3.5 | |
| Molmo 2-8Bparameters=8B2026.03 | 2.6 | 3.8 | |
| SAM 32026.03 | 2.5 | 2.9 | |
| GPT-4o2026.03 | 0.5 | 0.8 |