Video Reasoning on AoT Bench
68.9AccuracyOmniJigsaw (CMM)
Evaluation results
| Method | Links | |
|---|---|---|
| OmniJigsaw (CMM)Inference Mode=w/ Audio2026.04 | 68.9 | |
| OmniJigsaw (SMS)Inference Mode=w/ Audio2026.04 | 68.12 | |
| VideoJigsawInference Mode=w/ Audio2026.04 | 67.45 | |
| OmniJigsaw (JMI)Inference Mode=w/ Audio2026.04 | 66.83 | |
| OmniJigsaw (CMM)Inference Mode=w/o Audio2026.04 | 66.83 | |
| OmniJigsaw (SMS)Inference Mode=w/o Audio2026.04 | 66.39 | |
| VideoJigsawInference Mode=w/o Audio2026.04 | 66.22 | |
| OmniJigsaw (JMI)Inference Mode=w/o Audio2026.04 | 65.83 | |
| Qwen3-Omni-30BInference Mode=w/ Audio2026.04 | 64.88 | |
| Qwen3-Omni-30BInference Mode=w/o Audio2026.04 | 63.32 | |
| Video-R1Inference Mode=w/o Audio2026.04 | 52.6 | |
| Omni-R1Inference Mode=w/ Audio2026.04 | 52.09 | |
| Omni-R1Inference Mode=w/o Audio2026.04 | 51.03 | |
| HumanOmniV2Inference Mode=w/ Audio2026.04 | 48.58 | |
| HumanOmniV2Inference Mode=w/o Audio2026.04 | 47.35 |