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

Mathematical Visual Question Answering on MathVista

85.6Accuracy

Seed1.5-VL

Updated 9d ago

Evaluation Results

Method	Links
Seed1.5-VL 2026.03		85.6
Gemini-2.5-Pro 2026.03		82.7
Gemini 2.5 Pro 2026.04		82.7
OpenVLThinkerV2 2026.04		79.5
InternVL3.5-8B 2026.03		78.4
Qwen3-VL GRPO 2026.04		78.1
Qwen3-VL GDPO 2026.04		78
OneThinker-8B 2026.04		77.6
InternVL3.5-4B 2026.03		77.1
Vision-G1 2026.04		76.1
VL-Rethinker-7B 2026.04		74.9
ARES-7B 2026.04		74.6
Qwen3-VL-8B-Instruct 2026.03		74.2
Qwen3-VL-Instruct-8B 2026.04		74.2
Qwen2.5-VL-72B 2025.10		74
Qwen2.5-VL-72B + CARES 2025.10		74
VisionZero 2026.04		73.1
MM-Eureka-7B 2026.04		73
R-MSD (4B) 2026.03		72.1
OVR-7B 2026.04		72.1
InternVL3.5-2B 2026.03		71.8
InternVL3-8B 2026.03		71.6
Original SFT+RL (4B) 2026.03		71.2
OpenVLThinker-7B 2026.04		70.2
Qwen3-VL-4B-Instruct 2026.03		69.5
InternVL3-8B 2025.10		69
InternVL3-8B + CARES 2025.10		69
InternVL3-8B + CARES-AR 2025.10		68
VideoLLaMA3-7B 2026.03		67.1
InternVL2.5-8B 2026.03		64.4
GPT-4o 2025.10		64
GPT-4o 2026.03		63.8
GPT-4o 2026.04		63.8
R-MSD (2B) 2026.03		61.4
GPT-4o + CARES 2025.10		61
InternVL2.5-4B 2026.03		60.5
VideoLLaMA3-2B 2026.03		59.2
Qwen3-VL-2B-Instruct 2026.03		58.9
InternVL3-2B 2026.03		57
InternVL2.5-2B 2026.03		51.3
Granite-Vision -2B 2025.10		48
Granite-Vision -2B + CARES 2025.10		48
Granite-Vision -2B + CARES-AR 2025.10		44
SFT Vision 2025.10		38.4
MergeMix 2025.10		37.8
VisionThink-7B 2025.10		23.8
Qwen2.5-VL-Ins-7B 2025.10		19.2