List your company
ResearchFormerly WizWand

Complex Reasoning on Video-TT

46.5AccuracyOmniJigsaw (CMM)
19.4626.4833.540.52Mar 18, 2026Apr 3, 2026Apr 19, 2026May 5, 2026May 21, 2026Jun 6, 2026Jun 22, 2026
Updated 4mo ago

Evaluation results

MethodLinks
OmniJigsaw (CMM)Inference Mode=w/o Audio2026.0446.5
OmniJigsaw (CMM)Inference Mode=w/ Audio2026.0446.1
OmniJigsaw (SMS)Inference Mode=w/o Audio2026.0445.9
OmniJigsaw (SMS)Inference Mode=w/ Audio2026.0445.8
VideoJigsawInference Mode=w/ Audio2026.0444.9
VideoJigsawInference Mode=w/o Audio2026.0444.9
OmniJigsaw (JMI)Inference Mode=w/o Audio2026.0444.8
OmniJigsaw (JMI)Inference Mode=w/ Audio2026.0444.7
VideoLatent-7B (Ours)Frames=32, Architecture Category=Latent MLLM2026.0644.4
Qwen3-Omni-30BInference Mode=w/ Audio2026.0444.3
VideoLatent-7B (Ours)Frames=64, Architecture Category=Latent MLLM2026.0643.9
Qwen3-Omni-30BInference Mode=w/o Audio2026.0443.8
VideoLatent-7B (Ours)Frames=16, Architecture Category=Latent MLLM2026.0643.4
Video-R1-7BFrames=32, Architecture Category=Standard MLLM2026.0641.8
Video-R1-7BFrames=64, Architecture Category=Standard MLLM2026.0641.8
Qwen3-VL-8B + SynRLParameters=8B, Training=SynRL2026.0341.5
Mull-Token-7BFrames=16, Architecture Category=Latent MLLM2026.0641.3
Mull-Token-7BFrames=64, Architecture Category=Latent MLLM2026.0641.1
Video-R1-7BFrames=16, Architecture Category=Standard MLLM2026.0640.8
Mull-Token-7BFrames=32, Architecture Category=Latent MLLM2026.0640.8
Qwen3-VL-4B + SynRLParameters=4B, Training=SynRL2026.0340.7
Qwen3-VL-8BParameters=8B2026.0340.6
Video-R1Inference Mode=w/o Audio2026.0440.6
HumanOmniV2Inference Mode=w/ Audio2026.0440.3
Open-o3-Video-7BFrames=64, Architecture Category=Standard MLLM2026.0639.7
Open-o3-Video-7BFrames=32, Architecture Category=Standard MLLM2026.0639.3
LVR-7BFrames=64, Architecture Category=Latent MLLM2026.0639.3
Qwen3-VL-4BParameters=4B2026.0338.9
Qwen2.5-VL-7BFrames=64, Architecture Category=Standard MLLM2026.0638.7
LVR-7BFrames=32, Architecture Category=Latent MLLM2026.0638.3
HumanOmniV2Inference Mode=w/o Audio2026.0438.2
Qwen2.5-VL-7BFrames=32, Architecture Category=Standard MLLM2026.0637.9
Omni-R1Inference Mode=w/o Audio2026.0437
LVR-7BFrames=16, Architecture Category=Latent MLLM2026.0637
Qwen2.5-VL-7BFrames=16, Architecture Category=Standard MLLM2026.0636.9
Omni-R1Inference Mode=w/ Audio2026.0436.8
Monet-7BFrames=64, Architecture Category=Latent MLLM2026.0621.5
Monet-7BFrames=16, Architecture Category=Latent MLLM2026.0620.6
Monet-7BFrames=32, Architecture Category=Latent MLLM2026.0620.5