List your company
ResearchFormerly WizWand

Visual Logical Reasoning on LogicVista

70.02AccuracyCodePercept-32B-S1
30.895241.052651.2161.3674Sep 30, 2025Oct 27, 2025Nov 23, 2025Dec 20, 2025Jan 16, 2026Feb 12, 2026Mar 11, 2026
Updated 7mo ago

Evaluation results

MethodLinks
CodePercept-32B-S1LLM Solver=Qwen3-235A22-Thinking [53]2026.0370.02——
Gemini2.5-ProLLM Solver=Qwen3-30A3-Thinking [53]2026.0366.44——
Qwen3-VL-32B-InstructLLM Solver=Qwen3-235A22-Thinking [53]2026.0366.44——
CodePercept-32B-S1LLM Solver=Qwen3-30A3-Thinking [53]2026.0365.33——
CodePercept-4B-S1LLM Solver=Qwen3-235A22-Thinking [53]2026.0364.65——
CodePercept-8B-S1LLM Solver=Qwen3-235A22-Thinking [53]2026.0362.64——
Qwen3-VL-32B-InstructLLM Solver=Qwen3-30A3-Thinking [53]2026.0362.19——
CodePercept-8B-S1LLM Solver=Qwen3-30A3-Thinking [53]2026.0361.52——
CodePercept-4B-S1LLM Solver=Qwen3-30A3-Thinking [53]2026.0360.4——
Qwen3-VL-235A22B-InstructLLM Solver=Qwen3-30A3-Thinking [53]2026.0359.73——
Claude-Opus 4.1-ThinkingLLM Solver=Qwen3-30A3-Thinking [53]2026.0359.28——
Qwen3-VL-8B-InstructLLM Solver=Qwen3-235A22-Thinking [53]2026.0358.16——
Qwen3-VL-8B-InstructLLM Solver=Qwen3-30A3-Thinking [53]2026.0356.82——
Qwen3-VL-4B-InstructLLM Solver=Qwen3-235A22-Thinking [53]2026.0356.82——
Qwen3-VL-4B-InstructLLM Solver=Qwen3-30A3-Thinking [53]2026.0354.14——
Qwen3-VL-30A3B-InstructLLM Solver=Qwen3-30A3-Thinking [53]2026.0353.69——
GPT5-ThinkingLLM Solver=Qwen3-30A3-Thinking [53]2026.0353.02——
GPT4-oSampling Strategy=avg@8, Inference Temperature=12026.0152.8——
MiniCPM-V-4.5LLM Solver=Qwen3-30A3-Thinking [53]2026.0352.57——
Qwen2.5-VL-72BLLM Solver=Qwen3-30A3-Thinking [53]2026.0352.34——
Gemini-2.0-FlashSampling Strategy=avg@8, Inference Temperature=12026.0152.3——
InternVL3.5-8BLLM Solver=Qwen3-30A3-Thinking [53]2026.0351.68——
Intern-S1-8BLLM Solver=Qwen3-30A3-Thinking [53]2026.0351.23——
GLM-4.1V-9BLLM Solver=Qwen3-30A3-Thinking [53]2026.0351——
VAPOModel Scale=7B2025.0950.9——
RuCL2026.0249.66——
Qwen2.5-VL-32BParameters=32B2026.0249.26——
MM-EurekaModel Scale=7B2025.0948.5——
VisionZero-RealWorldModel Backbone=Qwen2.5-VL-7B-Instruct2026.0248.44——
Active-ZeroModel Backbone=Qwen2.5-VL-7B-Instruct2026.0248.21——
CPPOBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0148.2——
NoisyRolloutBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0147.9——
MM-Eureka-7BParameters=7B2026.0247.87——
GRPOBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0147.4——
PAPOModel Scale=7B, Decoding Strategy=greedy decoding2025.0947.3——
InternVL2.5-38BParameters=38B2026.0247.21——
VL-RethinkerModel Scale=7B2025.0947.2——
EvoLMMModel Backbone=Qwen2.5-VL-7B-Instruct2026.0246.65——
Look-BackBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0146.3——
VisionZero-ChartModel Backbone=Qwen2.5-VL-7B-Instruct2026.0246.21——
VL-Rethinker-7BParameters=7B2026.0246.08——
GPT-4o2026.0245.9——
OpenVLThinkerBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0145.8——
PAPOBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0145.8——
PerceptionR1Backbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0145.5——
KeyeVL1.5-8BLLM Solver=Qwen3-30A3-Thinking [53]2026.0345.19——
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0145.1——
Vision-MattersBackbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0145.1——
VisPlayModel Backbone=Qwen2.5-VL-7B-Instruct2026.0245.09——
R1-Onevision-7BParameters=7B2026.0244.53——
Claude-3.5-Sonnet2026.0243.97——
Vision-SR1Backbone=Qwen2.5-VL-7B, Sampling Strategy=avg@8, Inference Temperature=12026.0143.2——
Base ModelModel Backbone=Qwen2.5-VL-7B-Instruct2026.0243.08——
Vision-R1-7BParameters=7B2026.0242.94——
ThinkLite-VL-7BParameters=7B2026.0242.94——
Base ModelModel Scale=7B2025.0942.6——
VisionZero-CLEVRModel Backbone=Qwen2.5-VL-7B-Instruct2026.0241.29——
Perception-R1-7BParameters=7B2026.0240.94——
CPPOBackbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0140.9——
Active-ZeroModel Backbone=Qwen2.5-VL-3B-Instruct2026.0240.62——
PAPOBackbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0139.4——
VisPlayModel Backbone=Qwen2.5-VL-3B-Instruct2026.0239.29——
Qwen2.5-VL-7BParameters=7B2026.0239.26——
InternVL2.5-8BParameters=8B2026.0238.23——
GRPOBackbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0137.7——
OpenVLThinkerBackbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0137.4——
Base ModelModel Backbone=Qwen2.5-VL-3B-Instruct2026.0237.28——
Visionary-R1Backbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0137.1——
OpenVLThinker-7BParameters=7B2026.0236.24——
Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B, Sampling Strategy=avg@8, Inference Temperature=12026.0132.4——
InternVL2.5-8BParameters=8B2026.01—25.219
Janus-Pro-7BParameters=7B2026.01—20.427.9
LLaVA-OneVision-Qwen2-7BParameters=7B2026.01—20.625.3
MM-EurekaQwen-7BParameters=7B2026.01—34.622.8
Ours (RL)Training=RL2026.01—21.525.3
Ours (SFT)Training=SFT2026.01—2831.6
Qwen2.5-VL-7B-InstructParameters=7B, Variant=Instruct2026.01—25.219
R1-onevision-RLTraining=RL2026.01—2917.7
Thyme2026.01—33.619