mlsanityData to 14 Oct 2026

DeepSeek-OCR: Contexts Optical Compression

arXiv 2510.18234 ↗First posted 21 Oct 2025cs.CVcited by 183

Not on the map. Why?

Cited by 53 in Multimodal LLMs / vision-language models, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

We present DeepSeek-OCR as an initial investigation into the feasibility of compressing long contexts via optical 2D mapping. DeepSeek-OCR consists of two components: DeepEncoder and DeepSeek3B-MoE-A570M as the decoder. Specifically, DeepEncoder serves as the core engine, designed to maintain low activations under high-resolution input while achieving high compression ratios to ensure an optimal and manageable number of vision tokens. Experiments show that when the number of text tokens is within 10 times that of vision tokens (i.e., a compression ratio < 10x), the model can achieve decoding (OCR) precision of 97%. Even at a compression ratio of 20x, the OCR accuracy still remains at about 60%. This shows considerable promise for research areas such as historical long-context compression and memory forgetting mechanisms in LLMs. Beyond this, DeepSeek-OCR also demonstrates high practical value. On OmniDocBench, it surpasses GOT-OCR2.0 (256 tokens/page) using only 100 vision tokens, and outperforms MinerU2.0 (6000+ tokens per page on average) while utilizing fewer than 800 vision tokens. In production, DeepSeek-OCR can generate training data for LLMs/VLMs at a scale of 200k+ pages per day (a single A100-40G). Codes and model weights are publicly accessible at http://github.com/deepseek-ai/DeepSeek-OCR.

Key facts. Cited by 183 papers in the map (data through 14 Oct 2026). Builds on 30 of its 36 references that are in the map. First posted on arXiv 21 Oct 2025.

Builds on

30 of 36 references.

References · 30

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2023Segment Anythingcited by 8,234
  3. 2025Qwen2.5-VL Technical Reportcited by 5,587
  4. 2016SGDR: Stochastic Gradient Descent with Warm Restartscited by 5,045
  5. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  6. 2016Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answeringcited by 2,516
  7. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  8. 2019Towards VQA Models That Can Readcited by 1,718
  9. 2021LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairscited by 1,555
  10. 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoningcited by 1,479
  11. 2023MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilitiescited by 1,239
  12. 2024DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Modelcited by 1,135
18 more
  1. 2024How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suitescited by 1,050
  2. 2024DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understandingcited by 572
  3. 2024MinerU: An Open-Source Solution for Precise Document Content Extractioncited by 267
  4. 2023Nougat: Neural Optical Understanding for Academic Documentscited by 245
  5. 2022OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalizationcited by 236
  6. 2023Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolutioncited by 229
  7. 2024General OCR Theory: Towards OCR-2.0 via a Unified End-to-end Modelcited by 180
  8. 2024OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotationscited by 150
  9. 2025olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Modelscited by 125
  10. 2022Wukong: A 100 Million Large-scale Chinese Cross-modal Pre-training Benchmarkcited by 114
  11. 2025MonkeyOCR: Document Parsing with a Structure-Recognition-Relation Triplet Paradigmcited by 87
  12. 2025Dolphin: Document Image Parsing via Heterogeneous Anchor Promptingcited by 79
  13. 2024Focus Anywhere for Fine-grained Multi-page Document Understandingcited by 62
  14. 2025SmolDocling: An ultra-compact vision-language model for end-to-end multi-modal document conversioncited by 53
  15. 2024Small Language Model Meets with Reinforced Vision Vocabularycited by 43
  16. 2024OneChart: Purify the Chart Structural Extraction via One Auxiliary Tokencited by 39
  17. 2025PP-DocLayout: A Unified Document Layout Detection Model to Accelerate Large-Scale Data Constructioncited by 26
  18. 2024Slow Perception: Let's Perceive Geometric Figures Step-by-stepcited by 24

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 22; 2026: 161.

183 papers.

Sort:

By area:

  1. 2025Memory in the Age of AI Agentscited by 303
  2. 2026DeepSeek-OCR 2: Visual Causal Flowcited by 74
  3. 2026PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsingcited by 57
  4. 2025Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigmcited by 42
  5. 2026MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scalecited by 27
  6. 2026GLM-OCR Technical Reportcited by 26
  7. 2025Reasoning via Video: The First Evaluation of Video Models' Reasoning Abilities through Maze-Solving Taskscited by 22
  8. 2026AgentOCR: Reimagining Agent History via Optical Self-Compressioncited by 22
  9. 2026Dolphin-v2: Universal Document Parsing via Scalable Anchor Promptingcited by 22
  10. 2026Qianfan-OCR: A Unified End-to-End Model for Document Intelligencecited by 21
  11. 2025HunyuanOCR Technical Reportcited by 20
  12. 2025Context Cascade Compression: Exploring the Upper Limits of Text Compressioncited by 17
38 more
  1. 2026MemOCR: Layout-Aware Visual Memory for Efficient Long-Horizon Reasoningcited by 15
  2. 2026Multimodal OCR: Parse Anything from Documentscited by 14
  3. 2026HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Bettercited by 14
  4. 2025MonkeyOCR v1.5 Technical Report: Unlocking Robust Document Parsing for Complex Patternscited by 13
  5. 2025UniRec-0.1B: Unified Text and Formula Recognition with 0.1B Parameterscited by 13
  6. 2026FireRed-OCR Technical Reportcited by 13
  7. 2026OCRVerse: Towards Holistic OCR in End-to-End Vision-Language Modelscited by 12
  8. 2026EventMemAgent: Hierarchical Event-Centric Memory for Online Video Understanding with Adaptive Tool Usecited by 12
  9. 2026NextMem: Towards Latent Factual Memory for LLM-based Agentscited by 12
  10. 2025VTCBench: Can Vision-Language Models Understand Long Context with Vision-Text Compression?cited by 11
  11. 2026Youtu-Parsing: Perception, Structuring and Recognition via High-Parallelism Decodingcited by 10
  12. 2026ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thoughtcited by 10
  13. 2025DocPTBench: Benchmarking End-to-End Photographed Document Parsing and Translationcited by 9
  14. 2026VTC-R1: Vision-Text Compression for Efficient Long-Context Reasoningcited by 8
  15. 2026VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text?cited by 8
  16. 2026MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decodingcited by 8
  17. 2026UniVLR: Unifying Text and Vision in Visual Latent Reasoning for Multimodal LLMscited by 8
  18. 2026Unlimited OCR Workscited by 8
  19. 2025SpecPV: Improving Self-Speculative Decoding for Long-Context Generation via Partial Verificationcited by 7
  20. 2025Same Content, Different Answers: Cross-Modal Inconsistency in MLLMscited by 7
  21. 2026DeepRead: Document Structure-Aware Reasoning to Enhance Agentic Searchcited by 7
  22. 2026SPARC: Separating Perception And Reasoning Circuits for Test-time Scaling of VLMscited by 7
  23. 2026Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoderscited by 7
  24. 2026Towards Real-World Document Parsing via Realistic Scene Synthesis and Document-Aware Trainingcited by 7
  25. 2026HSD: Training-Free Acceleration for Document Parsing Vision-Language Models with Hierarchical Speculative Decodingcited by 6
  26. 2026Q-Zoom: Query-Aware Adaptive Perception for Efficient Multimodal Large Language Modelscited by 6
  27. 2026Beyond NL2Code: A Structured Survey of Multimodal Code Intelligencecited by 6
  28. 2025VRSA: Jailbreaking Multimodal Large Language Models through Visual Reasoning Sequential Attackcited by 5
  29. 2026FSVideo: Fast Speed Video Diffusion Model in a Highly-Compressed Latent Spacecited by 5
  30. 2026OmniScience: A Large-scale Multi-modal Dataset for Scientific Image Understandingcited by 5
  31. 2026AgenticOCR: Parsing Only What You Need for Efficient Retrieval-Augmented Generationcited by 5
  32. 2026Logics-Parsing-Omni Technical Reportcited by 5
  33. 2026Self-Improvement of Large Language Models: A Technical Overview and Future Outlookcited by 5
  34. 2026MDPBench: A Benchmark for Multilingual Document Parsing in Real-World Scenarioscited by 5
  35. 2026OvisOCR2 Technical Reportcited by 5
  36. 2025TRivia: Self-supervised Fine-tuning of Vision-Language Models for Table Recognitioncited by 4
  37. 2025InnoGym: Benchmarking the Innovation Potential of AI Agentscited by 4
  38. 2025Benchmarking Document Parsers on Mathematical Formula Extraction from PDFscited by 4

50 most cited of 183 All 183 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026Visual Merit or Linguistic Crutch? A Close Look at DeepSeek-OCR
  2. 2025Optical Context Compression Is Just (Bad) Autoencoding
  3. 2026RTPrune: Reading-Twice Inspired Token Pruning for Efficient DeepSeek-OCR Inference
  4. 2026Global Context Compression with Interleaved Vision-Text Transformation
  5. 2024DeepSeek-VL: Towards Real-World Vision-Language Understanding
  6. 2026Adversarial Attacks on Deep OCR Systems
  7. 2025VisionThink: Smart and Efficient Vision Language Model via Reinforcement Learning
  8. 2024FocusLLaVA: A Coarse-to-Fine Approach for Efficient and Effective Visual Token Compression