List your company

Video Reasoning on VBVR-Bench Out-of-Domain

61Average ScoreVBVR-Wan2.2
-2.1675214.2317430.63147.03026Mar 17, 2026Mar 29, 2026Apr 11, 2026Apr 24, 2026May 6, 2026May 19, 2026Jun 1, 2026

Evaluation results

MethodLinks
VBVR-Wan2.2Model Size=14B2026.056176.857.254.761.861.5
VideoRLVRModel Size=5B2026.0560.265.56259.758.858.2
Wan2.2-I2V-A14BModel Size=14B2026.0532.940.530.834.323.630.7
CogVideoX1.5Model Size=5B2026.0526.228.123.52525.428.2
HumanType=Baseline2026.030.988110.9910.97
VBVR-Wan2.2-5B-Distilled + VLM-as-TeacherCost (s)=692026.060.7590.8730.7650.7590.8180.639
VBVR-Wan2.2 + CollabVRVGM Cost (s)=10.912026.050.6960.8840.6340.6410.6080.72
VBVR-Wan2.2-5B + Pass@5Cost (s)=4352026.060.6610.7250.6640.6650.6950.564
VBVR-Wan2.2-5B + Pass@4Cost (s)=3482026.060.660.7230.6610.6650.6950.563
VBVR-Wan2.2-5B-Distilled + Pass@5Cost (s)=702026.060.6530.7090.6030.6510.7280.582
VBVR-Wan2.2-5B + Pass@3Cost (s)=2612026.060.6520.720.660.650.6820.56
VBVR-Wan2.2-5B-Distilled + Pass@4Cost (s)=562026.060.6520.7030.6030.6510.7270.581
VBVR-Wan2.2 + Training-Free EnsembleType=Video Reasoning Models, Ensemble=reasoning path ensembling2026.030.650.8030.7050.5310.6390.716
VBVR-Wan2.2-5B + Pass@2Cost (s)=1742026.060.650.7180.6590.6470.680.559
VBVR-Wan2.2-5B-Distilled + Pass@3Cost (s)=422026.060.650.7020.6030.6510.720.576
VBVR-Wan2.2-5B-Distilled + Pass@2Cost (s)=282026.060.6470.7010.6030.6510.7070.575
VBVR-Wan2.2-5B-DistilledCost (s)=142026.060.640.6880.6030.6510.6930.565
VBVR-Wan2.2-5BCost (s)=872026.060.6390.7110.6180.6420.6780.548
VBVR-Wan2.2-5B + VideoTPOCost (s)=2762026.060.6290.7030.6040.6310.6690.538
VBVR-Wan2.2 + Pass@4VGM Cost (s)=14.802026.050.6220.7850.660.5350.5770.683
VBVR-Wan2.2Type=Video Reasoning Models2026.030.610.7680.5720.5470.6180.615
VBVR-Wan2.2 + Pass@2VGM Cost (s)=7.402026.050.6020.7280.6170.4940.5320.701
VBVR-Wan2.2-14BCost (s)=1602026.060.6010.7320.5960.5420.6280.6
VBVR-Wan2.2-5B-Distilled + VideoTPOCost (s)=572026.060.5970.6520.5840.6130.6410.495
VBVR-Wan2.2 + VideoTPOVGM Cost (s)=11.102026.050.5820.7670.6190.5130.540.572
VBVR-Wan2.2VGM Cost (s)=3.702026.050.5770.6740.6740.5030.5280.633
Sora 2Type=Proprietary Video Models2026.030.5230.5460.4720.5250.4620.546
Sora 22026.060.5230.5460.4720.5250.4620.546
Veo 3.1Type=Proprietary Video Models2026.030.4290.5770.2770.420.4410.404
Veo 3.12026.060.4290.5770.2770.420.4410.404
Runway Gen-4 TurboType=Proprietary Video Models2026.030.4140.5150.4290.4190.3270.373
Cosmos-Predict2.5 + CollabVRVGM Cost (s)=10.912026.050.40.4810.2860.40.4710.346
Kling 2.6Type=Proprietary Video Models2026.030.330.5280.1350.2720.3560.359
Kling 2.62026.060.330.5280.1350.2720.3560.359
Wan2.2-I2V-A14BType=Open-source Video Models2026.030.3290.4050.3080.3430.2360.307
Cosmos-Predict2.5VGM Cost (s)=3.702026.050.3040.3680.1690.3090.3770.274
LTX-2Type=Open-source Video Models2026.030.2970.2440.3370.3170.2310.311
HunyuanVideo-I2VType=Open-source Video Models2026.030.2650.1750.3690.290.2530.25
CogVideoX1.5-5B-I2VType=Open-source Video Models2026.030.2620.2810.2350.250.2540.282