Our new X account is live! Follow @wizwand_team for updates

Visual Logical Reasoning on LogicVista

52.8Accuracy

GPT4-o

Updated 4d ago

Evaluation Results

Method	Links
GPT4-o 2026.01		52.8	-	-
Gemini-2.0-Flash 2026.01		52.3	-	-
VisionZero-RealWorld 2026.02		48.44	-	-
Active-Zero 2026.02		48.21	-	-
CPPO 2026.01		48.2	-	-
NoisyRollout 2026.01		47.9	-	-
GRPO 2026.01		47.4	-	-
EvoLMM 2026.02		46.65	-	-
Look-Back 2026.01		46.3	-	-
VisionZero-Chart 2026.02		46.21	-	-
OpenVLThinker 2026.01		45.8	-	-
PAPO 2026.01		45.8	-	-
PerceptionR1 2026.01		45.5	-	-
Qwen2.5-VL-7B 2026.01		45.1	-	-
Vision-Matters 2026.01		45.1	-	-
VisPlay 2026.02		45.09	-	-
Vision-SR1 2026.01		43.2	-	-
Base Model 2026.02		43.08	-	-
VisionZero-CLEVR 2026.02		41.29	-	-
CPPO 2026.01		40.9	-	-
Active-Zero 2026.02		40.62	-	-
PAPO 2026.01		39.4	-	-
VisPlay 2026.02		39.29	-	-
GRPO 2026.01		37.7	-	-
OpenVLThinker 2026.01		37.4	-	-
Base Model 2026.02		37.28	-	-
Visionary-R1 2026.01		37.1	-	-
Qwen2.5-VL-3B 2026.01		32.4	-	-
Qwen2.5-VL-7B-Instruct 2026.01		-	25.2	19
LLaVA-OneVision-Qwen2-7B 2026.01		-	20.6	25.3
InternVL2.5-8B 2026.01		-	25.2	19
Janus-Pro-7B 2026.01		-	20.4	27.9
R1-onevision-RL 2026.01		-	29	17.7
MM-EurekaQwen-7B 2026.01		-	34.6	22.8
Thyme 2026.01		-	33.6	19
Ours (SFT) 2026.01		-	28	31.6
Ours (RL) 2026.01		-	21.5	25.3