Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Visual Probing on VisualProbe Hard
Loading...
44.3
Accuracy
PERIA-8B
14.036
21.893
29.75
37.607
Jun 11, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
PERIA-8B
Model Type=Proposed
2026.06
44.3
Mini-o3
Model Type=Tool-Augmented
2026.06
41.5
Gemini-2.5-Flash
Model Type=Commercial
2026.06
33.6
Gemini-2.5-Pro
Model Type=Commercial
2026.06
32
w/o RL
Model Type=Ablation
2026.06
32
Qwen3-235B
Model Type=Open-Source
2026.06
27.3
Qwen3-8B
Model Type=Backbone Ba...
2026.06
26.6
w/o SFT
Model Type=Ablation
2026.06
26.4
GPT-5
Model Type=Commercial
2026.06
25.4
Qwen3-30B
Model Type=Open-Source
2026.06
25.4
Qwen3-32B
Model Type=Open-Source
2026.06
24.5
VTool-R1-7B
Model Type=Tool-Augmented
2026.06
22.6
InternVL3.5-14B
Model Type=Open-Source
2026.06
19.8
w/o Tool
Model Type=Ablation
2026.06
19.8
R1-OneVision-7B
Model Type=Tool-Augmented
2026.06
17.3
InternVL3.5-8B
Model Type=Open-Source
2026.06
15.2
Feedback
Search any
task
Search any
task