List your company
ResearchFormerly WizWand

Multimodal Mathematical Reasoning on LogicVista

75.2AccuracyGemini-2.5-Pro-Thinking
26.933639.464351.99564.5257Dec 19, 2025Jan 13, 2026Feb 7, 2026Mar 4, 2026Mar 29, 2026Apr 23, 2026May 19, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Gemini-2.5-Pro-ThinkingAccess Type=Closed-Source2025.1275.2
Seed-1.5-thinkingAccess Type=Closed-Source2025.1271.8
Claude-4-SonnetAccess Type=Closed-Source2025.1264.4
GPT-4.1Access Type=Closed-Source2025.1263.8
CEPOBackbone=Qwen3-VL-4B-Instruct2026.0561.16
RLSDBackbone=Qwen3-VL-4B-Instruct2026.0556.92
OPSDBackbone=Qwen3-VL-4B-Instruct2026.0555.58
GRPOBackbone=Qwen3-VL-4B-Instruct2026.0554.98
BaseBackbone=Qwen3-VL-4B-Instruct2026.0554.91
Revisual-R1Access Type=Open-Source, Model Scale=7B2025.1252.3
SDPOBackbone=Qwen3-VL-4B-Instruct2026.0552.01
VAPO-ThinkerAccess Type=Open-Source, Model Scale=7B2025.1250.9
MMR1-Math-v0Access Type=Open-Source, Model Scale=7B2025.1250.8
Vision-G1Access Type=Open-Source, Model Scale=7B2025.1250.2
OursAccess Type=Open-Source, Model Scale=7B2025.1249.9
Vision-R1Training Data=Mixed Data, Backbone=Qwen2.5-VL-7B2026.0349.7
RL(GRPO)Training Data=Geo3K, Backbone=Qwen2.5-VL-7B2026.0349.6
RL(GRPO)Training Data=MMR1, Backbone=Qwen2.5-VL-7B2026.0349.4
RL(GRPO)Training Data=GeoQA, Backbone=Qwen2.5-VL-7B2026.0349.2
VisionZeroTraining Data=ImgEdit, Backbone=Qwen2.5-VL-7B2026.0349.1
SelfJudgeTraining Data=Geo3K, Backbone=Qwen2.5-VL-7B2026.0349
VisionZeroTraining Data=CLEVR, Backbone=Qwen2.5-VL-7B2026.0348.8
SelfJudgeTraining Data=MMR1, Backbone=Qwen2.5-VL-7B2026.0348.6
VLAA-ThinkingAccess Type=Open-Source, Model Scale=7B2025.1248.5
SFTTraining Data=MMR1, Backbone=Qwen2.5-VL-7B2026.0348.1
MM-UPTTraining Data=MMR1, Backbone=Qwen2.5-VL-7B2026.0347.9
SelfJudgeTraining Data=GeoQA, Backbone=Qwen2.5-VL-7B2026.0347.9
SFTTraining Data=GeoQA, Backbone=Qwen2.5-VL-7B2026.0347.8
MM-UPTTraining Data=GeoQA, Backbone=Qwen2.5-VL-7B2026.0347.3
SFTTraining Data=Geo3K, Backbone=Qwen2.5-VL-7B2026.0347.2
EvoLMMTraining Data=Multi-Bench, Backbone=Qwen2.5-VL-7B2026.0346.9
MM-UPTTraining Data=Geo3K, Backbone=Qwen2.5-VL-7B2026.0346.9
MM-EurekaAccess Type=Open-Source, Model Scale=7B2025.1246.8
R1-Onevision-7BTraining Data=Mixed Data, Backbone=Qwen2.5-VL-7B2026.0346.5
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.0346.3
OpenVLThinker-8BTraining Data=Mixed Data, Backbone=Qwen2.5-VL-7B2026.0346.3
Qwen2.5-VLAccess Type=Open-Source, Model Scale=7B2025.1245
VL-RethinkerAccess Type=Open-Source, Model Scale=7B2025.1244.5
OpenVLThinkerAccess Type=Open-Source, Model Scale=7B2025.1241.4
ThinkLite-VLAccess Type=Open-Source, Model Scale=7B2025.1240.7
CEPOBackbone=Qwen3-VL-2B-Instruct2026.0537.72
GRPOBackbone=Qwen3-VL-2B-Instruct2026.0537.5
RLSDBackbone=Qwen3-VL-2B-Instruct2026.0536.38
BaseBackbone=Qwen3-VL-2B-Instruct2026.0532.81
SDPOBackbone=Qwen3-VL-2B-Instruct2026.0529.46
OPSDBackbone=Qwen3-VL-2B-Instruct2026.0528.79