Video Reasoning on TwiFF-Bench
3.11Avg CoTFUTURE-L1-RL
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FUTURE-L1-RLSize=8B2026.06 | 3.11 | — | — | — | — | — | — | 2.97 | 3.04 | |
| TwiFFSize=7B2026.02 | 2.95 | 2.81 | 2.43 | 3.24 | 3.04 | 3.32 | 3.14 | 2.62 | — | |
| TwiFF-2.7MSize=7B2026.06 | 2.95 | — | — | — | — | — | — | 2.62 | 2.79 | |
| TwiFF-LiteSize=7B2026.02 | 2.9 | 2.76 | 2.34 | 3.13 | 3.03 | 3.34 | 3.11 | 2.55 | — | |
| TwiFF-300KSize=7B2026.06 | 2.9 | — | — | — | — | — | — | 2.55 | 2.73 | |
| Qwen3VLSize=8B2026.02 | 2.84 | 2.78 | 2.31 | 2.88 | 2.8 | 3.1 | 2.76 | 2.44 | — | |
| Zero-ShotSize=8B2026.06 | 2.75 | — | — | — | — | — | — | 2.14 | 2.44 | |
| FUTURE-L1-SFTSize=8B2026.06 | 2.62 | — | — | — | — | — | — | 2.42 | 2.52 | |
| DeepEyesSize=7B2026.02 | 2.54 | 2.49 | 2.09 | 2.76 | 2.69 | 2.57 | 2.17 | 2.2 | — | |
| DeepEyesSize=7B2026.06 | 2.54 | — | — | — | — | — | — | 2.2 | 2.37 | |
| Qwen2.5VLSize=7B2026.02 | 2.46 | 2.4 | 1.64 | 2.57 | 1.77 | 2.64 | 1.41 | 1.63 | — | |
| Qwen2.5-VLSize=7B2026.06 | 2.46 | — | — | — | — | — | — | 1.63 | 2.05 | |
| InternVL3.5Size=8B2026.02 | 2.35 | 2.24 | 1.76 | 2.5 | 2.27 | 2.77 | 1.83 | 1.85 | — | |
| InternVL3.5Size=8B2026.06 | 2.35 | — | — | — | — | — | — | 1.85 | 2.1 | |
| BagelSize=7B2026.02 | 2.29 | 2.2 | 1.72 | 2.45 | 2.1 | 2.61 | 2.22 | 1.85 | — | |
| BagelSize=7B2026.06 | 2.29 | — | — | — | — | — | — | 1.85 | 2.07 | |
| Zebra-COTSize=7B2026.02 | 2.27 | 2.16 | 1.22 | 2.61 | 2.1 | 2.47 | 1.63 | 1.41 | — | |
| ThinkMorphSize=7B2026.02 | 2.21 | 2.15 | 1.3 | 2.48 | 2.38 | 2.2 | 1 | 1.43 | — | |
| Janus-ProSize=7B2026.02 | 2.04 | 1.95 | 1 | 2.14 | 1.36 | 2.38 | 0.93 | 1.04 | — | |
| Janus-ProSize=7B2026.06 | 2.04 | — | — | — | — | — | — | 1.04 | 1.54 |