List your company
ResearchFormerly WizWand

Vision-centric Reasoning on RealWorldQA

75.4AccuracyGPT-4o
43.36851.6846068.316Mar 4, 2025May 22, 2025Aug 10, 2025Oct 28, 2025Jan 16, 2026Apr 5, 2026Jun 24, 2026
Updated 4mo ago

Evaluation results

MethodLinks
GPT-4omodel_category=Proprietary Model2025.0975.4
SAPStrategy=SAP2026.0273.3
Gemini2.0-Flashmodel_category=Proprietary Model2025.0973.2
Qwen3-VL-8B-ThinkingStrategy=LongCoT2026.0272.8
MiMo-VL-7B-RLOpen Model=Yes, Open Data=No2025.1172.68
MiMo-VL-7B-SFTOpen Model=Yes, Open Data=No2025.1171.9
Qwen3-VL-8B-InstructStrategy=Instruct2026.0270.3
Long Grounded Thoughts (SFT + GRPO)Open Model=Yes, Open Data=Yes, Base Model=Qwen2.5-VL-7B-Instruct2025.1169.02
VisionZero-Qwen-7B (Real-World)training_context=Real-World, base_model=Qwen2.5-VL-7B2025.0968.9
Long Grounded Thoughts (Multistage SFT + DPO)Open Model=Yes, Open Data=Yes, Base Model=Qwen2.5-VL-7B-Instruct2025.1168.76
Grok-1.5V2026.0268.7
VisionZero-Qwen-7B (Chart)training_context=Chart, base_model=Qwen2.5-VL-7B2025.0968.5
VisionZero-Qwen-7B (CLEVR)training_context=CLEVR, base_model=Qwen2.5-VL-7B2025.0968.4
Qwen2.5-VL-7Bbase_model=Qwen2.5-VL-7B2025.0968.1
Qwen2.5-VL-7B-InstructOpen Model=Yes, Open Data=No2025.1167.84
Gemini-1.52026.0267.5
Qwen2.5-VL-7B-Instruct + LongPerceptualThoughtsOpen Model=Yes, Open Data=Yes2025.1167.45
Perceval#Param=7B2026.0467.4
DeepEyes#Param=7B, Tool Capability=true2026.0467
Qwen2.5-VL-7B-Instruct + VLAA-ThinkerOpen Model=Yes, Open Data=Yes2025.1166.93
ViGaL-Snake+Rotationbase_model=Qwen2.5-VL-7B2025.0966.5
OpenVLThinker#Param=7B2026.0466.5
Qwen2.5-VL + GRPO#Param=7B2026.0466.4
Long Grounded Thoughts (SFT + DPO)Open Model=Yes, Open Data=Yes, Base Model=Qwen2.5-VL-7B-Instruct2025.1166.14
MM-Eureka-Qwen-7Bbase_model=Qwen2.5-VL-7B2025.0966.1
Long Grounded Thoughts (SFT)Open Model=Yes, Open Data=Yes, Base Model=Qwen2.5-VL-7B-Instruct2025.1165.49
VLAA-Thinker-7Bbase_model=Qwen2.5-VL-7B2025.0965.4
Perceval#Param=3B2026.0464.9
Vision-R1#Param=7B2026.0464.1
Pixel-Reasoner#Param=7B, Tool Capability=true2026.0464
VL-Rethinker#Param=7B2026.0464
Qwen2.5-VL#Param=3B2026.0463.4
AKI-4BModel Scale=4B, Model Access Type=Open-source2025.0362.9
Qwen2.5-VL-7B-Instruct + Revisual-R1-finalOpen Model=Yes, Open Data=Yes2025.1162.48
MM-Eureka#Param=7B2026.0462.3
Qwen2.5-VL + GRPO#Param=3B2026.0462.1
VLAA-Thinker#Param=7B2026.0462
BLIP-3-4BModel Scale=4B, Model Access Type=Open-source2025.0361.6
VLM-R1#Param=3B2026.0461.5
GPT-4V2026.0261.4
VisRefModel=SAIL-VL2-8B2026.0261.2
Qwen2-VL-2BModel Scale=2B, Model Access Type=Open-source2025.0360.7
Qwen2.5-VL#Param=7B2026.0460.6
MolmoE-1B-7BAvg_k=8.002026.0660.5
OpenVLThinker-7Bbase_model=Qwen2.5-VL-7B2025.0960.2
GeMoEAvg_k=5.432026.0660.1
R1-VL#Param=7B2026.0459.5
VisRefModel=Qwen3-VL-8B2026.0259.1
Phi-3-Vision-4BModel Scale=4B, Model Access Type=Open-source2025.0358.8
LMM-R1#Param=3B2026.0458.7
TSRModel=SAIL-VL2-8B2026.0258
R1-OneVision-7Bbase_model=Qwen2.5-VL-7B2025.0958
Jigsaw-R1#Param=3B2026.0457.9
STModel=SAIL-VL2-8B2026.0257.3
TSRModel=Qwen3-VL-8B2026.0256.9
MM1.5-3BModel Scale=3B, Model Access Type=Proprietary2025.0356.9
Perception-R1#Param=3B2026.0455.8
MiniCPM-V2-3BModel Scale=3B, Model Access Type=Open-source2025.0355.8
STModel=Qwen3-VL-8B2026.0255.4
LLaVA-1.5-7BModel Scale=7B, Model Access Type=Open-source2025.0354.8
VILA-1.5-3BModel Scale=3B, Model Access Type=Open-source2025.0353.2
Claude-3-SonnetTable Header=Claude-S2026.0251.9
Claude-3-OpusTable Header=Claude-O2026.0249.8
DeepSeek-VL-1.3BModel Scale=1.3B, Model Access Type=Open-source2025.0349.7
R1-VL#Param=2B2026.0449.2
VisRefModel=InternVL3.5-8B2026.0247.2
TSRModel=InternVL3.5-8B2026.0244.9
STModel=InternVL3.5-8B2026.0244.6