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

Visual Mathematical Reasoning on WeMath

98.7Accuracy

OpenAI-o1

Updated 1mo ago

Evaluation Results

Method
OpenAI-o1 2026.02	98.7	-	-	-	-
o1 2026.06	98.7	-	-	-	-
Qwen3-VL-8B-Instruct + GSPO 2026.02	84	-	-	-	-
Octopus-8B (Ours) 2026.02	84	-	-	-	-
Qwen3-VL-8B-Thinking 2026.02	83	-	-	-	-
Qwen3-VL-8B-Instruct + DAPO 2026.02	82.4	-	-	-	-
Qwen3-VL-8B-Instruct + GRPO 2026.02	78.5	-	-	-	-
Gemini-2.5-Pro 2026.01	78	-	-	-	-
MiMo-VL-7B-SFT 2026.02	77.7	-	-	-	-
Qwen3-VL-8B-Instruct + SRPO 2026.02	76.9	-	-	-	-
Qwen3-VL-8B-Instruct + GRPO 2026.02	76.2	-	-	-	-
MiMo-VL-7B-RL 2026.02	76.1	-	-	-	-
Qwen3-VL-8B-Instruct + SRPO 2026.02	75.8	-	-	-	-
Qwen3-VL-8B-Instruct + GSPO 2026.02	74	-	-	-	-
Claude-3.7-Sonnet 2026.02	72.6	-	-	-	-
Claude-3.7-Sonnet 2025.08	72.6	-	-	-	-
Claude-3.7-Sonnet 2026.06	72.6	-	-	-	-
QvQ-72B-Preview 2026.06	72.4	-	-	-	-
Shuffle-R1-Qwen-7B 2025.08	72.3	-	-	-	-
Gemini-2.0-Flash 2026.06	71.4	-	-	-	-
GPT-5 2026.01	71.1	-	-	-	-
NoisyRollout-7B-K12 2025.08	70.8	-	-	-	-
MMR1-Math-7B 2025.08	70.7	-	-	-	-
Qwen3-VL-8B-Instruct 2026.02	70.5	-	-	-	-
R1-ShareVL 2026.06	70.17	-	-	-	-
VL-Rethinker-7B 2025.08	70.1	-	-	-	-
Active-Zero 2026.02	69.6	-	-	-	-
Qwen2.5-VL-7B-Instruct + VEPO 2026.06	69.54	-	-	-	-
VPPO-7B 2026.06	69.43	-	-	-	-
ThinkLite-VL-7B 2025.08	69.3	-	-	-	-
NoisyRollout-7B 2026.06	69.2	-	-	-	-
InternVL2.5-38B 2026.06	69.1	-	-	-	-
GPT-4o 2026.06	69	-	-	-	-
GPT-4o 2026.02	68.8	-	-	-	-
GLM-4.5V 2026.01	68.8	-	-	-	-
GPT-4o 2025.08	68.8	-	-	-	-
PAPO-DAPO-7B 2026.06	68.69	-	-	-	-
Qwen2.5-VL-7B-Instruct + GRPO (Top 40% high entropy) 2026.06	68.45	-	-	-	-
VisPlay 2026.02	67.99	-	-	-	-
VLAA-Thinker-7B 2025.08	67.7	-	-	-	-
Qwen2.5-VL-72B 2026.06	67.5	-	-	-	-
MM-Eureka-Qwen-7B 2025.08	67.4	-	-	-	-
OpenVLThinker-7B 2025.08	67.2	-	-	-	-
Qwen2.5-VL-7B-Instruct + GRPO 2026.06	67.18	-	-	-	-
CHARTOOL-7B 2026.04	67.13	-	-	-	-
Qwen2.5-VL-7B-Instruct + GRPO (Top 20% high entropy) 2026.06	67.13	-	-	-	-
VisionZero-CLEVR 2026.02	66.9	-	-	-	-
Shuffle-R1-Qwen-3B 2025.08	66.5	-	-	-	-
InternVL2.5-78B 2026.06	66.3	-	-	-	-
InternVL3.5-8B-RL 2026.02	65.8	-	-	-	-
Doubao-1.5-Pro 2026.01	65.7	-	-	-	-
VisionZero-RealWorld 2026.02	65.4	-	-	-	-
Qwen2.5-VL-32B 2026.06	65.4	-	-	-	-
Qwen2.5-VL-7B 2026.04	64.94	-	-	-	-
EvoLMM 2026.02	64.89	-	-	-	-
VisionZero-Chart 2026.02	64.66	-	-	-	-
Base Model 2026.02	64.48	-	-	-	-
COGFLOW 2026.01	64.1	-	-	-	-
GLM-4.1V 2026.01	63.8	-	-	-	-
Qwen2.5-VL-7B 2025.08	63.5	-	-	-	-
RL-Evol + Verifier (full VeriEvol) 2026.06	63.05	-	-	-	-
Qwen2.5-VL-3B-Instruct + VEPO 2026.06	62.7	-	-	-	-
VPPO-3B 2026.06	62.64	-	-	-	-
PAPO-DAPO-3B 2026.06	62.3	-	-	-	-
Qwen2.5-VL-7B-Instruct 2026.06	62.13	-	-	-	-
NoisyRollout-3B 2026.06	61.78	-	-	-	-
R1-ShareVL 2026.06	61.26	-	-	-	-
Active-Zero 2026.02	61.21	-	-	-	-
Qwen2.5-VL-3B-Instruct + GRPO 2026.06	61.15	-	-	-	-
Keye-VL 2026.01	60.7	-	-	-	-
R1-OneVision-7B 2025.08	60.6	-	-	-	-
Gemini2.5-Pro 2026.03	60.29	-	-	-	-
RL-Evol 2026.06	60	-	-	-	-
R1-VL-7B 2025.08	59.8	-	-	-	-
Qwen2.5-VL-3B-Instruct + GRPO (Top 40% high entropy) 2026.06	59.77	-	-	-	-
Qwen2.5-VL-3B-Instruct + GRPO (Top 20% high entropy) 2026.06	59.66	-	-	-	-
VeriEvol-SFT (VeriEvol-SFT-Init) 2026.06	59.33	-	-	-	-
CHARTOOL-3B 2026.04	58.45	-	-	-	-
RL-Origin 2026.06	58.19	-	-	-	-
VisPlay 2026.02	58.16	-	-	-	-
CodePercept-32B-S1 2026.03	57.81	-	-	-	-
InternVL3.5 2026.01	57	-	-	-	-
Seed-only SFT 2026.06	56.76	-	-	-	-
Qwen3-VL-32B-Instruct 2026.03	56.48	-	-	-	-
Qwen3-VL-8B 2026.05	56.1	-	-	-	-
Qwen3-VL-8B 2026.05	56.1	-	-	-	-
Qwen3-VL-8B 2026.05	55.43	-	-	-	-
GPT5-Thinking 2026.03	54.57	-	-	-	-
OneThinker-8B 2026.05	54.57	-	-	-	-
CodePercept-32B-S1 2026.03	54.19	-	-	-	-
Qwen2.5-VL-3B-Instruct 2026.06	53.97	-	-	-	-
OpenMMReasoner-7B 2026.06	53.81	-	-	-	-
Base Model 2026.02	53.51	-	-	-	-
Qwen3-VL-235A22B-Instruct 2026.03	53.05	-	-	-	-
CodePercept-8B-S1 2026.03	52.29	-	-	-	-
Qwen2.5-VL-3B 2025.08	51.7	-	-	-	-
Qwen3-VL-8B 2026.05	50.86	-	-	-	-
Qwen3-VL-8B 2026.05	50.86	-	-	-	-
PRCO-7B 2026.03	50.29	-	-	-	-
CodePercept-4B-S1 2026.03	50	-	-	-	-

Showing 100 of 208 rows