Video Reasoning on TUNA-Bench
66.2AccuracyOmniJigsaw (CMM)
Evaluation Results
| Method | Links | |
|---|---|---|
| OmniJigsaw (CMM)Inference Mode=w/o Audio2026.04 | 66.2 | |
| OmniJigsaw (CMM)Inference Mode=w/ Audio2026.04 | 65.29 | |
| OmniJigsaw (SMS)Inference Mode=w/ Audio2026.04 | 65.15 | |
| OmniJigsaw (SMS)Inference Mode=w/o Audio2026.04 | 64.73 | |
| VideoJigsawInference Mode=w/o Audio2026.04 | 64.39 | |
| OmniJigsaw (JMI)Inference Mode=w/o Audio2026.04 | 63.9 | |
| Qwen3-Omni-30BInference Mode=w/o Audio2026.04 | 63.62 | |
| VideoJigsawInference Mode=w/ Audio2026.04 | 63.41 | |
| OmniJigsaw (JMI)Inference Mode=w/ Audio2026.04 | 62.78 | |
| Qwen3-Omni-30BInference Mode=w/ Audio2026.04 | 62.57 | |
| Video-R1Inference Mode=w/o Audio2026.04 | 55.94 | |
| Omni-R1Inference Mode=w/ Audio2026.04 | 53.84 | |
| Omni-R1Inference Mode=w/o Audio2026.04 | 52.65 | |
| HumanOmniV2Inference Mode=w/o Audio2026.04 | 49.44 | |
| HumanOmniV2Inference Mode=w/ Audio2026.04 | 49.16 |