List your company
ResearchFormerly WizWand

Multimodal Reasoning on MathVista (Accuracy)

83.5AccuracyInternVL2.5-38B + VRPRM
19.74836.29952.8569.401Jun 11, 2025Aug 14, 2025Oct 17, 2025Dec 20, 2025Feb 22, 2026Apr 27, 2026Jul 1, 2026
Updated 4mo ago

Evaluation results

MethodLinks
InternVL2.5-38B + VRPRMBase Model=InternVL2.5-38B, Selection Strategy=Best-of-8, Critic Model=VRPRM2025.0883.5
InternVL2.5-26B + VRPRMBase Model=InternVL2.5-26B, Selection Strategy=Best-of-8, Critic Model=VRPRM2025.0881.2
AnE-3rdTraining Stage=Round 32026.0581.2
AnE-2ndTraining Stage=Round 22026.0580.1
AnE-1stTraining Stage=Round 12026.0579.6
OpenMMReasoner-7BEvaluation Source=original paper2026.0579.5
InternVL2.5-8B + VRPRMBase Model=InternVL2.5-8B, Selection Strategy=Best-of-8, Critic Model=VRPRM2025.0879.1
Athena-PRMPolicy Model=Qwen2.5-VL-72B, Best-of-N Evaluation=82025.0679.1
InternVL2.5-38B + VRPRM w/o RLBase Model=InternVL2.5-38B, Selection Strategy=Best-of-8, Critic Model=VRPRM w/o RL2025.0878.4
Preliminary RLTraining Phase=Warm-up2026.0577.9
Athena-ORMPolicy Model=Qwen2.5-VL-72B, Best-of-N Evaluation=82025.0677.8
InternVL2.5-26B + VRPRM w/o RLBase Model=InternVL2.5-26B, Selection Strategy=Best-of-8, Critic Model=VRPRM w/o RL2025.0877.4
Self-consistencyPolicy Model=Qwen2.5-VL-72B, Best-of-N Evaluation=82025.0677
SPARK-VL-7BEvaluation Source=original paper2026.0575.9
Metis-RISEEvaluation Source=original paper2026.0575.8
SRPOEvaluation Source=original paper2026.0575.8
Athena-PRMPolicy Model=Qwen2.5-VL-7B, Best-of-N Evaluation=82025.0675.2
VL-RethinkerEvaluation Source=original paper2026.0574.9
ADHintEvaluation Source=original paper2026.0574.4
Qwen2.5-VL-72BPolicy Model=Qwen2.5-VL-72B, Best-of-N Evaluation=82025.0674.2
LLaVA-Critic-R1Evaluation Source=original paper2026.0574
InternVL2.5-38B + VisualPRMBase Model=InternVL2.5-38B, Selection Strategy=Best-of-8, Critic Model=VisualPRM2025.0873.9
Vision-R1Evaluation Source=reproduced2026.0573.5
InternVL2.5-26B + VisualPRMBase Model=InternVL2.5-26B, Selection Strategy=Best-of-8, Critic Model=VisualPRM2025.0873.1
Revisual-R1Evaluation Source=original paper2026.0573.1
Athena-ORMPolicy Model=Qwen2.5-VL-7B, Best-of-N Evaluation=82025.0672.8
InternVL2.5-8B + VRPRM w/o RLBase Model=InternVL2.5-8B, Selection Strategy=Best-of-8, Critic Model=VRPRM w/o RL2025.0872.6
OpenVLThinkerEvaluation Source=original paper2026.0572.3
MMR1Evaluation Source=original paper2026.0572
InternVL2.5-38BBase Model=InternVL2.5-38B, Selection Strategy=Single-shot2025.0871.9
Self-consistencyPolicy Model=Qwen2.5-VL-7B, Best-of-N Evaluation=82025.0671.6
Athena-PRMPolicy Model=InternVL2.5-8B, Best-of-N Evaluation=82025.0671.4
Gemini-2.0-FlashSelection Strategy=Standard2025.0870.4
VisualPRM-8BPolicy Model=Qwen2.5-VL-7B, Best-of-N Evaluation=82025.0670.3
C2-EvoEvaluation Source=original paper2026.0568.7
InternVL2.5-8B + VisualPRMBase Model=InternVL2.5-8B, Selection Strategy=Best-of-8, Critic Model=VisualPRM2025.0868.5
VisualPRM-8BPolicy Model=InternVL2.5-8B, Best-of-N Evaluation=82025.0668.5
InternVL2.5-26BBase Model=InternVL2.5-26B, Selection Strategy=Single-shot2025.0868.2
Qwen2.5-VL-7B-InstructEvaluation Source=original paper2026.0568.2
Qwen2.5-VL-7BPolicy Model=Qwen2.5-VL-7B, Best-of-N Evaluation=82025.0668.1
Athena-ORMPolicy Model=InternVL2.5-8B, Best-of-N Evaluation=82025.0666.9
Self-consistencyPolicy Model=InternVL2.5-8B, Best-of-N Evaluation=82025.0666.1
Claude-3.5-SonnetSelection Strategy=Standard2025.0865.3
InternVL2.5-8BBase Model=InternVL2.5-8B, Selection Strategy=Single-shot2025.0864.5
InternVL2.5-8BPolicy Model=InternVL2.5-8B, Best-of-N Evaluation=82025.0664.5
GPT-4oSelection Strategy=Standard2025.0860
BaselineModel=Qwen2 [92]2026.0756.1
ESCModel=Qwen2 [92]2026.0756.1
ESCModel=LLaVA [50]2026.0723.4
BaselineModel=LLaVA [50]2026.0722.2