List your company
ResearchFormerly WizWand

Multimodal Understanding on MME Perception

1,699.5MME-P ScoreVLM-UniDDT
-21.388425.381872.151,318.919Oct 17, 2024Jan 26, 2025May 7, 2025Aug 16, 2025Nov 25, 2025Mar 6, 2026Jun 15, 2026
Updated 4mo ago

Evaluation results

MethodLinks
VLM-UniDDT# Params.=4B+1B2026.061,699.5——
Qwen2.5 VLParams=7B, Architecture category=Unifying Understanding and Generation via Next-token Prediction, VLMEvalKit=true2026.031,692.5——
WallarooParams=7B, Architecture category=Unifying Understanding and Generation via Next-token Prediction, VLMEvalKit=true2026.031,690.3——
BAGELParams=7B+7B, Architecture category=Integrating Autoregressive and Diffusion within Transformers2026.031,687——
BAGEL# Params.=14B2026.061,687——
MetaQueryParams=7B + 1.6B, Architecture category=Multi-modal Understanding Models as Enhanced Conditional Encoders2026.031,685.2——
Blip3-oParams=7B + 1.4B, Architecture category=Multi-modal Understanding Models as Enhanced Conditional Encoders2026.031,682.6——
Show-o2Params=7B, Architecture category=Integrating Autoregressive and Diffusion within Transformers2026.031,620.5——
Show-o2# Params.=7B2026.061,620.5——
MogaoParams=7B, Architecture category=Integrating Autoregressive and Diffusion within Transformers2026.031,592——
Mogao# Params.=7B2026.061,592——
LLaVA-OneVisionLLM=Qwen2-7B2025.051,580——
LLaVA-OV# Params.=7B2026.061,580——
ShareGPT4VLLM=Vicuna-7B2025.051,567.4——
Janus-Pro-7BType=Und. and Gen., LLM Params=7B2025.011,567.1——
Janus-ProParams=7B, Architecture category=Unifying Understanding and Generation via Next-token Prediction2026.031,567.1——
Janus-Pro# Params.=7B2026.061,567.1——
TokenFlow-XL*# Params.=14B2026.061,551.1——
TokenFlow-XLType=Und. and Gen., LLM Params=13B2025.011,545.9——
TokenFlowParams=13B, Architecture category=Unifying Understanding and Generation via Next-token Prediction2026.031,545.9——
LLaVA-NeXTLLM=Vicuna-7B2025.051,519——
LLaVA-v1.5Type=Und. Only, # LLM Params=7B, External pretrained diffusion model=false2024.101,510.7——
LLaVA-v1.5Type=Und. Only, LLM Params=7B2025.011,510.7——
LLaVA-v1.5# Params.=7B2026.061,510.7——
Qwen-VL# Params.=7B2026.061,487.6——
Qwen-VL-ChatType=Und. Only, # LLM Params=7B, External pretrained diffusion model=false2024.101,487.5——
Qwen-VL-ChatType=Und. Only, LLM Params=7B2025.011,487.5——
MorphTokensLLM=Vicuna-7B, PT Data Size=30M2025.051,477.7——
Slot-MLLM (14B)LLM=Qwen2.5-14B-Instruct, PT Data Size=25M2025.051,468.3——
Slot-MLLM (14B)LLM=Qwen2.5-14B-Instruct2025.051,468.3——
UniTokLLM=LLaMA-2-7B, PT Data Size=70M2025.051,465.7——
Liquid# Params.=8B2026.061,448——
ILLUMEType=Und. and Gen., LLM Params=7B2025.011,445.3——
ILLUME# Params.=7B2026.061,445.3——
Janus-Pro-1BType=Und. and Gen., LLM Params=1.5B2025.011,444——
Janus-Pro# Params.=1.5B2026.061,444——
MobileVLM-V2Type=Und. Only, # LLM Params=2.7B, External pretrained diffusion model=false2024.101,440.5——
MobileVLM-V2Type=Und. Only, LLM Params=2.7B2025.011,440.5——
VILA-UType=Und. and Gen., # LLM Params=7B, External pretrained diffusion model=false2024.101,401.8——
VILA-UType=Und. and Gen., LLM Params=7B2025.011,401.8——
VILA-UParams=7B, Architecture category=Unifying Understanding and Generation via Next-token Prediction2026.031,401.8——
VILA-U# Params.=7B2026.061,401.8——
JanusType=Und. and Gen., # LLM Params=1.3B, External pretrained diffusion model=false2024.101,338——
JanusType=Und. and Gen., LLM Params=1.5B2025.011,338——
JanusParams=1.5B, Architecture category=Unifying Understanding and Generation via Next-token Prediction2026.031,338——
VILA-ULLM=LLaMA-2-7B, PT Data Size=23M2025.051,336.2——
LLaVA-PhiType=Und. Only, # LLM Params=2.7B, External pretrained diffusion model=false2024.101,335.1——
LLaVA-PhiType=Und. Only, LLM Params=2.7B2025.011,335.1——
JanusFlowParams=1.3B, Architecture category=Integrating Autoregressive and Diffusion within Transformers2026.031,333.1——
JanusFlow# Params.=1.5B2026.061,333.1——
MobileVLM-V2Type=Und. Only, # LLM Params=1.4B, External pretrained diffusion model=false2024.101,302.8——
MobileVLM-V2Type=Und. Only, LLM Params=1.4B2025.011,302.8——
MobileVLMType=Und. Only, # LLM Params=2.7B, External pretrained diffusion model=false2024.101,288.9——
MobileVLMType=Und. Only, LLM Params=2.7B2025.011,288.9——
Emu3-ChatType=Und. Only, LLM Params=8B2025.011,244——
Slot-MLLM (7B)LLM=Vicuna-7B, PT Data Size=25M2025.051,225.2——
Slot-MLLM (7B)LLM=Vicuna-7B2025.051,225.2——
InstructBLIPType=Und. Only, # LLM Params=13B, External pretrained diffusion model=false2024.101,212.8——
InstructBLIPType=Und. Only, LLM Params=13B2025.011,212.8——
MobileVLMType=Und. Only, # LLM Params=1.4B, External pretrained diffusion model=false2024.101,196.2——
MobileVLMType=Und. Only, LLM Params=1.4B2025.011,196.2——
LLaVA-v1.5-Phi-1.5Type=Und. Only, # LLM Params=1.3B, External pretrained diffusion model=false2024.101,128——
LLaVA-v1.5-Phi-1.5Type=Und. Only, LLM Params=1.3B2025.011,128——
D-DitType=Und. and Gen., LLM Params=2.0B2025.011,124.7——
SEED-LLaMALLM=Vicuna-7B, PT Data Size=77M2025.051,123.9——
Show-oType=Und. and Gen., LLM Params=1.3B, Resolution=5122025.011,097.2——
Show-oParams=1.3B, Architecture category=Integrating Autoregressive and Diffusion within Transformers2026.031,097.2——
Show-o# Params.=1.3B2026.061,097.2——
LaVITLLM=LLaMA-7B, PT Data Size=100M2025.05964.5——
Show-oType=Und. and Gen., # LLM Params=1.3B, External pretrained diffusion model=false2024.10948.4——
Show-oType=Und. and Gen., LLM Params=1.3B, Resolution=2562025.01948.4——
LLAVAType=Und. Only, # LLM Params=7B, External pretrained diffusion model=false2024.10809.6——
LLaVAType=Und. Only, LLM Params=7B2025.01809.6——
LWMLLM=LLaMA-2-7B, PT Data Size=-2025.0544.8——
AKI-4BModel Scale=4B, Model Access Type=Open-source2025.03—1,491.9—
Bagel#LLM=14B2025.12——1,687
BLIP-2zero-shot=true2023.05—290—
BLIP-3-4BModel Scale=4B, Model Access Type=Open-source2025.03—1,487.6—
BLIP3-o#LLM=8B2025.12——1,682.6
Bunnynumber of parameters=8B2025.07——1,987.7
Cambrian-1Size=8B, Type=AR, Samples=-2025.12—1,547—
ChatVLAnumber of parameters=1.5B2025.07——1,435.2
COINCIDE2025.02—1,495.6—
DataTailor2025.02—1,476.1—
DeepSeek-VL-1.3BModel Scale=1.3B, Model Access Type=Open-source2025.03—1,306.6—
DiffusionVLSize=3B, Type=Diff., Samples=738K2025.12—1,539—
DiffusionVLSize=7B, Type=Diff., Samples=738K2025.12—1,519—
DimpleSize=7B, Type=Diff., Samples=1.3M2025.12—1,514—
Eagle2number of parameters=1.5B2025.07——1,572.1
ECoTnumber of parameters=7B2025.07——0
EL2N2025.02—1,356.5—
EMU3#LLM=8B2025.12——1,243.8
Full-Finetune2025.02—1,510.7—
GraNd2025.02—1,400.5—
Honeybee-C-7BModel Scale=7B, Model Access Type=Open-source2025.03—1,584.2—
ICONS2025.02—1,485.7—
ILLUME+#LLM=3B2025.12——1,414
InstructionGPT-42025.02—463.3—
InstructVLA-Generalistnumber of parameters=1.5B2025.07——1,529.6
InstructVLA-Generalistnumber of parameters=1.5B, robot state=true2025.07——1,548