DeepSeek-OCR

Headlines · 3

1. 3d
   Apple's LensVLM compresses long context into images, then zooms in

   Apple's foundation-model team and Duke University propose LensVLM, which first lets a model skim compressed page images that cover the full text, then calls a tool to expand the pages relevant to the question and read the original content. The paper reports 68.9% average accuracy across seven text QA benchmarks at 4.3x effective compression, close to the 72.4% full-text reference; under 15x initial compression the base model scores 18.1% versus 52.1% for LensVLM. Built on Qwen3.5-9B-Base, it uses Qwen3.5-397B to generate tool-use trajectories for supervised fine-tuning, followed by DAPO reinforcement learning; the work was accepted to NeurIPS 2026 and a LensVLM 9B demo is online.

   机器之心 · 🔥 0
2. 3d
   Baidu open-sources Unlimited-OCR, a 3B long-horizon document parsing model

   Baidu published Unlimited-OCR on Hugging Face, a 3B image-text-to-text model under an MIT license that the card frames as "one-shot long-horizon parsing," aiming to push beyond DeepSeek-OCR. It supports single images in two configs (gundam and base) plus multi-page and PDF parsing with a 32,768 max length and no-repeat n-gram constraints, and the card ships instructions for Transformers, vLLM and SGLang inference, ms-swift training, a ModelScope mirror, Baidu Cloud availability and a Hugging Face Spaces demo. The release timeline runs from June 22, 2026, the paper is listed as arXiv 2606.23050, and the authors credit DeepSeek-OCR, DeepSeek-OCR-2 and PaddleOCR; post-processing code for OmniDocBench evaluation is also provided.

   百度文心 · Hugging Face · 🔥 0
3. Oct 8
   Ai2 releases olmOCR v0.4.0, lifting benchmark score ~4 points with RL

   The Allen Institute for AI released olmOCR v0.4.0 on October 21, 2025, a new model trained with synthetic data and reinforcement learning that raises the project's olmOCR-Bench score by roughly 4 points. The toolkit converts PDF, PNG and JPEG documents into clean Markdown, handling equations, tables, handwriting, multi-column layouts and insets while stripping headers and footers and preserving natural reading order. Ai2 says it runs on a 7B vision-language model requiring a GPU, at under $200 per million pages, and reports an olmOCR-Bench score of 82.4±1.1 for v0.4.0 — just below Chandra OCR 0.1.0 (83.1) on its own leaderboard.

   GitHub Trending · Python · 🔥 0

Experience and discussion from the community

Share my DeepSeek-OCR experienceAsk about DeepSeek-OCR