Unlimited OCR 2 is a vision-language model built around extracting and interpreting text from images. It handles long documents with a large context window, making it capable of processing multi-page scanned materials in a single pass. As an open-weight model under the MIT license, it can be deployed and fine-tuned freely.