List your company

Visual Question Answering on DocVQA (ANLS)

95.01ANLSOScaR
8.263630.784353.30575.8257Oct 7, 2022May 21, 2023Jan 2, 2024Aug 15, 2024Mar 29, 2025Nov 10, 2025Jun 25, 2026

Evaluation results

MethodLinks
OScaRBackbone=Qwen3-VL-8B, Bit-width=INT2, Group Size=1282026.0595.01
16bitBackbone=Qwen3-VL-8B, Bit-width=16-bit2026.0594.93
OTTBackbone=Qwen3-VL-8B, Bit-width=INT2, Group Size=1282026.0594.86
KIVIBackbone=Qwen3-VL-8B, Bit-width=INT2, Group Size=1282026.0594.7
TurboQuant+Backbone=Qwen3-VL-8B, Bit-width=2.5-bit2026.0594.51
16bitBackbone=Qwen3-VL-4B, Bit-width=16-bit2026.0594.23
QuaRotBackbone=Qwen3-VL-8B, Bit-width=INT2, Group Size=1282026.0593.93
OScaRBackbone=Qwen3-VL-4B, Bit-width=INT2, Group Size=1282026.0593.85
InternVL-2.5-26B*Latency=0.74542025.1293.78
TurboQuant+Backbone=Qwen3-VL-4B, Bit-width=2.5-bit2026.0593.54
KIVIBackbone=Qwen3-VL-4B, Bit-width=INT2, Group Size=1282026.0593.49
OTTBackbone=Qwen3-VL-4B, Bit-width=INT2, Group Size=1282026.0593.24
RotateKVBackbone=Qwen3-VL-8B, Bit-width=INT2, Group Size=1282026.0592.76
LUVCBackbone=InternVL-2.5-26B*, Latency=0.4302, Speedup=1.732025.1292.7
InternVL-2.5-8B*Latency=0.29852025.1291.96
LUVCBackbone=InternVL-2.5-8B*, Latency=0.1857, Speedup=1.612025.1291.48
QuaRotBackbone=Qwen3-VL-4B, Bit-width=INT2, Group Size=1282026.0589.89
ViQBase LLM=Qwen2.5-7B, Size=1.3B, AnyRes=true, Discrete=true2026.0688.9
VTWBackbone=InternVL-2.5-26B*, Latency=0.4637, Speedup=1.612025.1288.78
PaLI-3OCR pipeline input=true, Resolution=1064x10642023.1088.6
ERNIE-ViL 2.0OCR pipeline input=true, Resolution=1064x10642023.1088.4
PaLI-3OCR pipeline input=false, Resolution=1064x10642023.1087.6
TILTInput type=Pipelined, Additional labeled data=true2022.1087.1
PACTBackbone=InternVL-2.5-26B*, Latency=0.4690, Speedup=1.592025.1285.79
InternViT-2.5-6BBase LLM=Qwen2.5-7B, Size=6.0B, AnyRes=false, Discrete=false2026.0685.7
UDOPInput type=Pipelined2022.1084.7
PACTBackbone=InternVL-2.5-8B*, Latency=0.1980, Speedup=1.512025.1284.32
ViQBase LLM=Qwen2.5-1.5B, Size=1.3B, AnyRes=true, Discrete=true2026.0684.2
LayoutLMv3Input type=Pipelined2022.1083.4
InternViT-2.5Base LLM=Qwen2.5-1.5B, Size=0.3B, AnyRes=false, Discrete=false2026.0681.7
T5 + 2D + UInput type=Pipelined2022.1081
VTWBackbone=InternVL-2.5-8B*, Latency=0.1804, Speedup=1.652025.1280.59
LeoPT=595 K, SFT=1 M2025.0180.1
InternViT-2.5-6BBase LLM=Qwen2.5-1.5B, Size=6.0B, AnyRes=false, Discrete=false2026.0680.1
AIMv2-HBase LLM=Qwen2.5-7B, Size=0.7B, AnyRes=false, Discrete=false2026.0680.1
PaLI-XOCR pipeline input=false, Resolution=1064x10642023.1080
PALI-X-55BModel type=Specialist SOTAs, Configuration=Single-task fine-tuning, without OCR Pipeline, Parameters=55B2023.0880
OryxViTBase LLM=Qwen2.5-7B, Size=0.4B, AnyRes=true, Discrete=false2026.0679.8
Eagle-X2PT=595 K, SFT=1.8 M2025.0177.7
Eagle-X3PT=595 K, SFT=1.8 M2025.0177.7
DinoV2-gBase LLM=Qwen2.5-1.5B, Size=1.1B, AnyRes=false, Discrete=false2026.0676.9
Pix2Struct-LargeInput type=Pixel-only, Scale=Large2022.1076.6
Pix2Struct-Large (1.3B)Model type=Generalist Models, Parameters=1.3B2023.0876.6
Pix2StructResolution=10242024.0476.6
RotateKVBackbone=Qwen3-VL-4B, Bit-width=INT2, Group Size=1282026.0576.12
SigLIP2-gBase LLM=Qwen2.5-7B, Size=1.1B, AnyRes=false, Discrete=false2026.0675
AIMv2-HBase LLM=Qwen2.5-1.5B, Size=0.7B, AnyRes=false, Discrete=false2026.0673.7
OryxViTBase LLM=Qwen2.5-1.5B, Size=0.4B, AnyRes=true, Discrete=false2026.0673.5
Pix2Struct-BaseInput type=Pixel-only, Scale=Base2022.1072.1
HRVDAResolution=15362024.0472.1
SeRumResolution=1280, Fine-tuning=true2024.0471.9
ROBERTaInput type=Pipelined2022.1069.5
16bitBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=16-bit2026.0568.57
SigLIP2-gBase LLM=Qwen2.5-1.5B, Size=1.1B, AnyRes=false, Discrete=false2026.0667.8
OScaRBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=INT2, Group Size=1282026.0567.79
DonutInput type=Pixel-only2022.1067.5
DonutResolution=1280, Fine-tuning=true2024.0467.5
TurboQuant+Backbone=LLaVA-v1.6-vicuna-7B, Bit-width=2.5-bit2026.0567.09
OTTBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=INT2, Group Size=1282026.0566.11
QuaRotBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=INT2, Group Size=1282026.0565.55
UReaderResolution=2242024.0465.4
KIVIBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=INT2, Group Size=1282026.0565.34
RotateKVBackbone=LLaVA-v1.6-vicuna-7B, Bit-width=INT2, Group Size=1282026.0565.26
Qwen-VL (Qwen-7B)Model type=Generalist Models, Backbone=Qwen-7B2023.0865.1
Qwen-VLResolution=448, Open-source evaluation=true2024.0465.1
Qwen-VL-ChatModel type=Generalist Models2023.0862.6
mPLUG-DocOwl (LLaMA-7B)Model type=Generalist Models, Backbone=LLaMA-7B2023.0862.2
mPLUG-DocResolution=2242024.0462.2
OAI-CLIP-LBase LLM=Qwen2.5-7B, Size=0.3B, AnyRes=false, Discrete=false2026.0661.4
OAI-CLIP-LBase LLM=Qwen2.5-1.5B, Size=0.3B, AnyRes=false, Discrete=false2026.0652.9
MoME-X3PT=NA, SFT=NA2025.0150.8
SCVMBackbone=LLaVA-1.5-7B2026.0221
LLaVA-1.5-7BBackbone=LLaVA-1.5-7B2026.0217
QLIPBase LLM=Qwen2.5-1.5B, Size=0.3B, AnyRes=false, Discrete=true2026.0612.2
UniTokBase LLM=Qwen2.5-1.5B, Size=0.3B, AnyRes=false, Discrete=true2026.0611.6