List your company

Long-context document understanding on MMLongBench-Doc

55.8AccuracySynthetic Reasoning
4.42417.76231.144.438Oct 8, 2024Jan 8, 2025Apr 10, 2025Jul 12, 2025Oct 12, 2025Jan 12, 2026Apr 15, 2026

Evaluation results

MethodLinks
Synthetic ReasoningModel family=Qwen3 VL, Scale=32B, Strategy=Synthetic Reasoning2026.0355.8—
Plain DistillationModel family=Qwen3 VL, Strategy=Plain Distillation2026.0354.9—
Qwen3 VL 235B A22B InstructModel family=Qwen3 VL, Scale=235B A22B, Strategy=Instruct2026.0354.8—
LongPOModel family=Qwen3 VL, Strategy=LongPO2026.0353.6—
No-thinkModel family=Qwen3 VL, Strategy=No-think2026.0352.2—
Qwen3 VL 32B InstructModel family=Qwen3 VL, Scale=32B, Strategy=Instruct2026.0351.8—
Synthetic ReasoningModel family=Mistral, Scale=24B, Strategy=Synthetic Reasoning2026.0347.5—
Plain DistillationModel family=Mistral, Strategy=Plain Distillation2026.0346.6—
GPT-4.1Backbone=-, Param=-, Paradigm=E2E2026.0445.6—
Qwen Thinking TracesModel family=Mistral, Strategy=Thinking Traces2026.0345—
No-thinkModel family=Mistral, Strategy=No-think2026.0343—
GPT-4ocategory=Proprietary2024.1042.944.9
GPT-4oVenue=-, Param.=-, Evi.=P, RAG=×2026.0442.8—
GPT-4oBackbone=-, Param=-, Paradigm=E2E2026.0442.8—
Doc-V* (GRPO)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0442.1—
MoLoRAGBackbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.0441—
DocSeekerVenue=-, Param.=7B, Evi.=P, RAG=×2026.0440.1—
Mistral 3.1 Small 24BModel family=Mistral, Scale=24B2026.0339.9—
Doc-V* (SFT)Backbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0439.8—
DocSeeker-SFTVenue=-, Param.=7B, Evi.=P, RAG=×2026.0438.6—
Qwen2.5-VL (RAG Top-5)Backbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.0436.1—
Claude-3.7-SonnetBackbone=-, Param=-, Paradigm=E2E2026.0433.9—
URaGBackbone=Qwen2.5-VL, Param=7B, Paradigm=RAG2026.0433.8—
Qwen2-VL-72Bactivated_params=72B, category=Open-source2024.1033.335.7
CogDocBackbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0433—
GPT-4VVenue=-, Param.=-, Evi.=P, RAG=×2026.0432.4—
GPT-4o minicategory=Proprietary2024.102928.6
DocopilotVenue=CVPR 2025, Param.=8B, Evi.=T+I, RAG=×2026.0428.8—
Baseline-SFT (short-answer)Venue=-, Param.=7B, Evi.=P, RAG=×2026.0428.8—
DocopilotBackbone=InternVL2, Param=8B, Paradigm=E2E2026.0428.8—
GPT-4o miniBackbone=-, Param=-, Paradigm=E2E2026.0428.6—
ARIAactivated_params=3.9B (25.3B), category=Open-source2024.1028.324.6
Gemini-1.5-Procategory=Proprietary2024.1028.220.6
Gemini-1.5-ProVenue=-, Param.=-, Evi.=P, RAG=×2026.0428.2—
Gemini-1.5-ProBackbone=-, Param=-, Paradigm=E2E2026.0428.2—
Qwen2.5-VL (Baseline)Backbone=Qwen2.5-VL, Param=7B, Paradigm=E2E2026.0428—
Gemini-1.5-Flashcategory=Proprietary2024.102721.3
VRAG-RLBackbone=Qwen2.5-VL, Param=7B, Paradigm=Agent2026.0426.6—
BaselineVenue=-, Param.=7B, Evi.=P, RAG=×2026.0425.4—
InternVL3Venue=Arxiv 2025, Param.=8B, Evi.=P, RAG=×2026.0424.1—
InternVL3Backbone=InternViT / Qwen2.5, Param=8B, Paradigm=E2E2026.0424.1—
SV-RAGVenue=ICLR 2025, Param.=4B, Evi.=P, RAG=✓2026.0423—
SV-RAGBackbone=InternVL2, Param=4B, Paradigm=RAG2026.0423—
Qwen2-VL-7Bactivated_params=7B, category=Open-source2024.1021.322.7
M3DocRAGVenue=Arxiv 2024, Param.=9B, Evi.=P, RAG=✓2026.0421—
M3DocRAGBackbone=Qwen2-VL, Param=7B, Paradigm=RAG2026.0421—
Vis-RAGVenue=ICLR 2025, Param.=4B, Evi.=P, RAG=✓2026.0418.8—
VisRAGBackbone=MiniCPM-V 2.6, Param=8B, Paradigm=RAG2026.0418.8—
VDocRAGVenue=CVPR 2025, Param.=8B, Evi.=P, RAG=✓2026.0418.4—
VDocRAGBackbone=Phi3-Vision, Param=4B, Paradigm=RAG2026.0418.4—
InternVL2-40Bactivated_params=40B, category=Open-source2024.1018.217.9
InternVL-Chat-V1.5activated_params=26B, category=Open-source2024.1014.613
Llama3.2-11Bactivated_params=11B, category=Open-source2024.1013.811.3
mPLUG-DocOwl2Venue=ACL 2025, Param.=8B, Evi.=P, RAG=×2026.0413.4—
mPLUG-DocOwl2Backbone=ViT / LLaMa, Param=8B, Paradigm=E2E2026.0413.4—
MiniCPM-V-2.6activated_params=8B, category=Open-source2024.1011.511.6
Idefics2activated_params=8B, category=Open-source2024.1076.8
Pixtral-12Bactivated_params=12B, category=Open-source2024.106.46