Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

V*Bench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Visual Question AnsweringV*Bench
Accuracy98.95
94
Visual ReasoningV*Bench
Accuracy95.7
62
Visual Perception and ReasoningV* Bench
Attribute Score98.3
49
Visual SearchV* Bench
Accuracy90.4
41
Visually Grounded ReasoningV* Bench
Average Accuracy95.7
32
Visual Perception ReasoningV* Bench
Score89.01
28
Fine-grained Visual Question AnsweringV*Bench
Overall Accuracy92.15
28
Vision-Centric ReasoningV* Bench (Overall)
Attribute Score96.5
24
High-Resolution Image PerceptionV* Bench
Overall Score91.6
22
Visual perception of small objectsV* Bench
Accuracy94.76
19
Real-World UnderstandingV* Bench
Accuracy85.6
18
Visual UnderstandingV* Bench
Avg@8 EM0.942
18
Fine-grained visual understandingV* Bench
General Score85.5
18
Visually Grounded ReasoningV* Bench (test)
Overall Accuracy95
17
Visual GroundingV* Bench
Overall Success Rate95.7
17
Multimodal ReasoningV* Bench Tool-needed
Accuracy90.1
15
Visual Perception and ReasoningV* Bench 1.0 (test)
Attribute Score83.48
13
Fine-grained PerceptionV*Bench
Pass@187.4
10
High-Resolution PerceptionV*-Bench v1.0 (test)
Overall Score83.8
10
Fine-grained Visual PerceptionV* Bench
Overall Score95.7
10
Visual PerceptionV*Bench
Accuracy84.3
9
Visual Tool-UseV* Bench
Accuracy88.2
9
Fine-grained Visual PerceptionV* Bench (OOD)
Accuracy76.4
6
Multimodal Question AnsweringV* Bench
Answer Accuracy80.6
4
Text-to-Video GenerationV-Bench
Generation Speed (x)3.2
4
Showing 25 of 27 rows