List your company

Visual Number Reasoning on VisNum

47.94AccuracyActive-Zero
34.180837.752941.32544.8971Feb 11, 2026Feb 28, 2026Mar 17, 2026Apr 3, 2026Apr 20, 2026May 7, 2026May 24, 2026

Evaluation results

MethodLinks
Active-ZeroBackbone=Qwen2.5-VL-7B-Instruct2026.0247.94
VisPlayBackbone=Qwen2.5-VL-7B-Instruct2026.0245.22
EvoLMMBackbone=Qwen2.5-VL-7B-Instruct2026.0245.01
Base ModelBackbone=Qwen2.5-VL-7B-Instruct2026.0244.85
VisionZero-ChartBackbone=Qwen2.5-VL-7B-Instruct2026.0244.54
VisionZero-RealWorldBackbone=Qwen2.5-VL-7B-Instruct2026.0244.17
VisionZero-CLEVRBackbone=Qwen2.5-VL-7B-Instruct2026.0243.13
DUELRole=Solver2026.0542.7
Vision-Zero(CLEVR)2026.0542.3
OpenVLThinker-7B2026.0542.2
EvoLMM2026.0541.9
DUELRole=Challenger2026.0541.9
VLAA-Thinker-7B2026.0541.8
Qwen2.5-VL-7B2026.0541.7
MM-UPT2026.0541.6
Active-ZeroBackbone=Qwen2.5-VL-3B-Instruct2026.0241.51
VisPlayBackbone=Qwen2.5-VL-3B-Instruct2026.0237.32
Base ModelBackbone=Qwen2.5-VL-3B-Instruct2026.0234.71