List your company
ResearchFormerly WizWand

General VQA on MMStar

80.1AccuracyQwen3.5-Plus
31.2243.9156.669.29Mar 18, 2026Apr 4, 2026Apr 22, 2026May 9, 2026May 27, 2026Jun 13, 2026Jul 1, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen3.5-PlusParameter Count=397B2026.0680.1
K2.5Parameter Count=≈1T2026.0679.8
Qwen3-VL-32BDecoding Mode=Thinking, Model Scale=32B2026.0677.1
Qwen3-VL-32BDecoding Mode=Instruct, Model Scale=32B2026.0676
Qwen3-VL-8BDecoding Mode=Thinking, Model Scale=8B2026.0675.2
Opus-4.6Parameter Count=≈800B2026.0674.5
Jigsaw-R1-8BModel Category=Template-based Self-evolution Methods, Scale=8B2026.0474.3
EVE (Ours-8B-iter4)Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=42026.0473.9
VisPlay-8B-iter3Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=32026.0473.1
GPT-5.4Parameter Count=≈1.5T2026.0672.7
Qwen3-VL-8B-InstructModel Category=Open-Source MLLMs, Scale=8B2026.0472.1
Yuvion VL-32BDecoding Mode=Reasoning, Model Scale=32B2026.0671.3
Yuvion VL-32BDecoding Mode=Instruct, Model Scale=32B2026.0671
Qwen2.5-VL-7B AIFmethod=Adaptive Information Flow2026.0470.9
Qwen2.5-VL-72B2026.0470.8
MM-Zero-8B-iter3Model Category=Pseudo-label-based Self-evolution Methods, Scale=8B, Iteration=32026.0470.7
Yuvion VL-8BDecoding Mode=Reasoning, Model Scale=8B2026.0670.5
Qwen3-VL-8BDecoding Mode=Instruct, Model Scale=8B2026.0670.1
Qwen3-VLModel Size=4B, Training Stage=instruct2026.0569.8
M2NoteBackbone=Qwen3-VL-Plus, Evolving Mode=Self-Evolving, Mem=28, Len=4622026.0769.4
GPT-5 nano (high)Model Category=Closed-Source MLLMs, Scale=high2026.0468.6
VanillaBackbone=Qwen3-VL-Plus, Evolving Mode=Self-Evolving2026.0768.3
M2NoteBackbone=GPT-5.4, Evolving Mode=Self-Evolving, Mem=28, Len=4622026.0767.8
Qwen3-VL-SegModel Size=4B, Training Stage=S-22026.0567.7
Qwen3-VLModel Size=4B, Training Stage=S-12026.0567.5
VanillaBackbone=GPT-5.4, Evolving Mode=Self-Evolving2026.0766.5
InternVL3-9BModel Category=Open-Source MLLMs, Scale=9B2026.0466.3
Qwen-ViPER-7BModel Category=Pseudo-label-based Self-evolution Methods, Scale=7B2026.0466.2
LLaVA-OneVision-72BModel Category=Open-Source MLLMs, Scale=72B2026.0465.8
Yuvion VL-8BDecoding Mode=Instruct, Model Scale=8B2026.0665.6
Vision-ZeroEvolving Mode=Self-Evolving2026.0765.2
Claude-3.5 Sonnet2026.0465.1
VisPlayEvolving Mode=Self-Evolving2026.0765.1
InternVL-3.5Model Size=4B2026.0565
CoT+M2NoteBackbone=Qwen3-VL-8B-Instruct, Evolving Mode=Self-Evolving, Mem=32, Len=3152026.0764.9
GPT-4o2026.0464.7
GPT-4o-20240513Model Category=Closed-Source MLLMs2026.0464.7
M2NoteBackbone (Tuning)=Qwen3-VL-8B-Instruct, Backbone (Tuner)=Qwen3-VL-Plus, Evolving Mode=Cross-Model Evolving, Mem=43, Len=4452026.0764
Qwen2.5-VL-7B2026.0463.9
M2NoteBackbone=Qwen3-VL-8B-Instruct, Evolving Mode=Self-Evolving, Mem=50, Len=2612026.0763.9
CoTBackbone=Qwen3-VL-8B-Instruct, Evolving Mode=Self-Evolving2026.0762.9
VanillaBackbone=Qwen3-VL-8B-Instruct, Evolving Mode=Self-Evolving2026.0762.1
DPEBackbone=Qwen3-VL-8B-Instruct, Evolving Mode=Self-Evolving2026.0762.1
VanillaBackbone (Tuning)=Qwen3-VL-8B-Instruct, Backbone (Tuner)=Qwen3-VL-Plus, Evolving Mode=Cross-Model Evolving2026.0762.1
GPT-5 mini (minimal)Model Category=Closed-Source MLLMs, Scale=minimal2026.0461.3
FineViT-VLLanguage Backbone=Qwen3 1.7B LLM, Evaluation Framework=VLMEvalKit [69]2026.0360.87
Intern3.5-VLLanguage Backbone=Qwen3 1.7B LLM, Evaluation Framework=VLMEvalKit [69]2026.0357.33
Qwen3-VLLanguage Backbone=Qwen3 1.7B LLM, Evaluation Framework=VLMEvalKit [69]2026.0356.6
Aquila-VLLanguage Backbone=Qwen2.5 1.5B LLM, Evaluation Framework=VLMEvalKit [69]2026.0354.53
LLaVA-1.5-7B AIFmethod=Adaptive Information Flow2026.0439.5
LLaVA-1.5-7B2026.0433.1