List your company

Multimodal Mathematical Reasoning on MathVista

85.1GPS ScoreClaude-4-Sonnet
33.146.660.173.6Feb 4, 2026

Evaluation results

MethodLinks
Claude-4-SonnetModel Category=Proprietary2026.0285.181.9
Gemini-2.5-ProModel Category=Proprietary2026.0279.3379
Guided-GRPOBackbone=Qwen3-8B, Training=G-GRPO+SFT2026.0277.8876.51
Vision-R1-32BModel Category=Open-Source2026.027574.73
Qwen2.5-VL-32BModel Category=Open-Source2026.0270.6761.57
GPT-4oModel Category=Proprietary2026.0269.2367.97
Qwen-VL-MaxModel Category=Proprietary2026.0268.7558.36
Base (Qwen3-8B)Backbone=Qwen3-8B2026.0267.3164.06
VL-Rethinker-7BModel Category=Open-Source2026.0266.9365.12
InternVL2.5-26BModel Category=Open-Source2026.0265.3853.38
Llama-3.2-11BModel Category=Open-Source2026.0260.5855.16
LLaVA-v1.5-13BModel Category=Open-Source2026.0257.6951.6
MMR1-32BModel Category=Open-Source2026.0243.7544.48
Math-LLaVA-13BModel Category=Open-Source2026.0242.7939.86
Phi-3-vision-128kModel Category=Open-Source2026.0235.133.45