Jina released Jina-OCR-v1, an end-to-end document parsing model. It uses a compressed-vision encoder and a 3B mixture-of-experts decoder from DeepSeek-OCR. A FastMTP speculative decoding head shares a single draft block across three prediction steps. Greedy verification keeps decoding lossless. Post-training combines instruction alignment, robustness fine-tuning, and GRPO under dense verifiable rewards. The model scores 91.14 on OmniDocBench v1.6 and 83.4 on olmOCR-Bench. It reaches 2.57 pages per second throughput. FastMTP doubles decoding speed over greedy autoregressive decoding on an NVIDIA L4. The model is publicly available on Hugging Face.
What changed
Greedy autoregressive decoding was the default method for document parsing.
What it unlocks
Running document parsing on low-budget GPUs at double the decoding speed.
- 91.14 OmniDocBench v1.6 score
- 83.4 olmOCR-Bench score
- 2.57 pages per second throughput
- 570M parameters per token
What you need to act on it
- a low-budget GPU such as NVIDIA L4
Sources