mlsanityData to 14 Oct 2026

HunyuanOCR-1.5: Making Lightweight OCR VLMs Faster and Better

arXiv 2607.04884 ↗First posted 6 Jul 2026cs.CVcited by 14

On the map: Multimodal LLMs / vision-language models → Document understanding and multimodal RAG

Abstract (arXiv)

We present HunyuanOCR-1.5, a lightweight end-to-end OCR-specialized vision-language model. HunyuanOCR unifies document parsing, text spotting, information extraction, text-image translation, and multi-image document understanding within a single end-to-end VLM. Building upon the lightweight architecture of HunyuanOCR-1.0, HunyuanOCR-1.5 does not redesign the backbone, but systematically improves both efficiency and capability. For efficiency, we adapt DFlash to OCR decoding, significantly reducing the latency of long structured outputs such as dense documents, tables, and formulas while preserving output distribution. Powered by DFlash, HunyuanOCR-1.5 achieves a 6.37x Transformer inference speedup and a 2.14x speedup under vLLM, delivering the fastest inference among lightweight OCR VLMs. For capability, we propose Agentic Data Flow, an agent-driven data construction system that transforms model weaknesses into executable data requirements and autonomously performs material search, quality verification, and pipeline development. It substantially improves long-tail capabilities in ancient-script OCR, fine-grained chart and table parsing, multi-image text-centric QA, low-resource multilingual parsing, and document hallucination evaluation. HunyuanOCR-1.5 ranks among the top-tier end-to-end OCR solutions on OmniDocBench v1.6 while achieving new performance milestones across these long-tail tasks. Combined with an upgraded pretraining and post-training recipe, HunyuanOCR-1.5 further extends its capability in high-resolution, long-context, and multi-task scenarios. Experiments demonstrate faster inference, broader OCR capability coverage, and the deployment advantages of a lightweight end-to-end model. We will release the model weights and training code to support future research and real-world OCR applications.

Key facts. Cited by 14 papers in the map (data through 14 Oct 2026). Placed in Document understanding and multimodal RAG, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 104 of its 120 references that are in the map. First posted on arXiv 6 Jul 2026.

Builds on

104 of 120 references.

Closest ancestors · 12

Same direction, newest first

  1. 2026ABot-OCR Technical Report27 May
  2. 2026Qianfan-OCR: A Unified End-to-End Model for Document Intelligence11 Mar
  3. 2026HSD: Training-Free Acceleration for Document Parsing Vision-Language Models with Hierarchical Speculative Decoding13 Feb
  4. 2026OCRVerse: Towards Holistic OCR in End-to-End Vision-Language Models29 Jan
  5. 2025dots.ocr: Multilingual Document Layout Parsing in a Single Vision-Language Model2 Dec
  6. 2025Infinity Parser: Layout Aware Reinforcement Learning for Scanned Document Parsing17 Oct
  7. 2025PaddleOCR-VL: Boosting Multilingual Document Parsing via a 0.9B Ultra-Compact Vision-Language Model16 Oct
  8. 2025MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing26 Sep
  9. 2025Logics-Parsing Technical Report24 Sep
  10. 2025POINTS-Reader: Distillation-Free Adaptation of Vision-Language Models for Document Conversion1 Sep
  11. 2025olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models25 Feb
  12. 2023OCRBench: On the Hidden Mystery of OCR in Large Multimodal Models13 May

Other references · 92

  1. 2017Attention Is All You Needcited by 60,361
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  4. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  5. 2023GPT-4 Technical Reportcited by 18,396
  6. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  7. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  8. 2025Qwen2.5-VL Technical Reportcited by 5,587
  9. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  10. 2024GPT-4o System Cardcited by 4,588
  11. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  12. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,796
38 more
  1. 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
  2. 2025DAPO: An Open-Source LLM Reinforcement Learning System at Scalecited by 2,810
  3. 2025Qwen3-VL Technical Reportcited by 2,793
  4. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
  5. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,224
  6. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  7. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,653
  8. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,599
  9. 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiencycited by 1,510
  10. 2024How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suitescited by 1,050
  11. 2025OpenAI GPT-5 System Cardcited by 989
  12. 2024Gemma: Open Models Based on Gemini Research and Technologycited by 916
  13. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 843
  14. 2025DeepSeek-V3.2: Pushing the Frontier of Open Large Language Modelscited by 815
  15. 2023PyTorch FSDP: Experiences on Scaling Fully Sharded Data Parallelcited by 660
  16. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertaintycited by 578
  17. 2026Kimi K2.5: Visual Agentic Intelligencecited by 501
  18. 2025GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learningcited by 466
  19. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Modelscited by 409
  20. 2021OCR-free Document Understanding Transformercited by 407
  21. 2016Synthetic Data for Text Localisation in Natural Imagescited by 396
  22. 2024EAGLE-2: Faster Inference of Language Models with Dynamic Draft Treescited by 370
  23. 2024Better & Faster Large Language Models via Multi-token Predictioncited by 360
  24. 2025EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Testcited by 331
  25. 2025Reinforcement Learning with Verifiable Rewards Implicitly Incentivizes Correct Reasoning in Base LLMscited by 289
  26. 2024Unlocking Efficiency in Large Language Model Inference: A Comprehensive Survey of Speculative Decodingcited by 269
  27. 2026Gemma 4 Technical Reportcited by 259
  28. 2024Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimizationcited by 239
  29. 2025Seg-Zero: Reasoning-Chain Guided Segmentation via Cognitive Reinforcementcited by 239
  30. 2019Image-based table recognition: data, model, and evaluationcited by 237
  31. 2023Patch n' Pack: NaViT, a Vision Transformer for any Aspect Ratio and Resolutioncited by 229
  32. 2025MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipecited by 186
  33. 2024Flex Attention: A Programming Model for Generating Optimized Attention Kernelscited by 185
  34. 2024MMLongBench-Doc: Benchmarking Long-context Document Understanding with Visualizationscited by 183
  35. 2025DeepSeek-OCR: Contexts Optical Compressioncited by 183
  36. 2024OmniDocBench: Benchmarking Diverse PDF Document Parsing with Comprehensive Annotationscited by 150
  37. 2025PaddleOCR 3.0 Technical Reportcited by 143
  38. 2025Fast-dLLM v2: Efficient Block-Diffusion LLMcited by 139

50 most cited of 92 All 92 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 14.

14 papers.

Sort:

By area:

  1. 2026OvisOCR2 Technical Reportcited by 5
  2. 2026MonkeyOCRv2: A Visual-Text Foundation Model for Document AIcited by 4
  3. 2026PaDoc: Layout-Grounded Parallel Decoding for Document Parsingcited by 1
  4. 2026HPD-Parsing: Hierarchical Parallel Document Parsingcited by 0
  5. 2026TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documentscited by 0
  6. 2026WildHandBench: A Benchmark for Handwritten Text Understanding that Challenges MLLMs and Humanscited by 0
  7. 2026Jina-OCR-v1: Efficient Document Parsing with Speculative Decoding and Dense Verifiable Rewardscited by 0
  8. 2026WeVisDoc: From Coverage to Capability for Robust End-to-End Document Parsingcited by 0
  9. 2026Diffusion Drafts, AR Verifies: Accelerating Document OCR with Self-Speculative Decodingcited by 0
  10. 2026When VLMs Trust Context: Evaluating Scene Text Recognition under Misleading Contextcited by 0
  11. 2026Xiaomi-OCR-0 Technical Reportcited by 0
  12. 2026Decompose Radicals, Then Reward: Fine-Grained Inspection for Accurate Chinese Text Renderingcited by 0
2 more
  1. 2026PolyOCR-Venus: Unified OCR Foundation Models for Text-Centric Visual Intelligencecited by 0
  2. 2026Improving OCR Faithfulness via Gated and Attenuated On-Policy Distillationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025HunyuanOCR Technical Report
  2. 2026PP-OCRv5: A Specialized 5M-Parameter Model Rivaling Billion-Parameter Vision-Language Models on OCR Tasks
  3. 2026Do VLMs Read or Rewrite? On Transcription Faithfulness in Vision-Language Models
  4. 2025DianJin-OCR-R1: Enhancing OCR Capabilities via a Reasoning-and-Tool Interleaved Vision-Language Model
  5. 2026PP-OCRv6: From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks
  6. 2026LightOnOCR: A 1B End-to-End Multilingual Vision-Language Model for State-of-the-Art OCR
  7. 2026DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines
  8. 2026Where Vision Becomes Text: Locating the OCR Routing Bottleneck in Vision-Language Models