Jina OCR V1
Jina AI🇩🇪Major lab
TextVisionopen
Jina OCR V1 is a document parsing model built on DeepSeek-OCR that uses FastMTP speculative decoding for faster page transcription. It scores 91.14 on OmniDocBench v1.6 and 83.4 on olmOCR-Bench.
Releases from frontier labs, major companies and research institutes — grouped by model, deduplicated across HuggingFace, OpenRouter and official blogs.
Click a week to filter
Jina AI🇩🇪Major lab
Jina OCR V1 is a document parsing model built on DeepSeek-OCR that uses FastMTP speculative decoding for faster page transcription. It scores 91.14 on OmniDocBench v1.6 and 83.4 on olmOCR-Bench.