OCR models / OpenDataLab (Shanghai AI Lab)
MinerU-Diffusion V1
OpenDataLab (Shanghai AI Lab) · released 2026-03-24 · data checked 2026-11-06
MinerU-Diffusion V1 is a 2.7B-parameter OCR and document-parsing model from OpenDataLab (Shanghai AI Lab) (CN), released 2026-03-24 under the MIT license. It has no published score on OmniDocBench v1.6 or olmOCR-Bench yet. It needs about 6.4 GB of memory at 16-bit.
Experimental diffusion decoder that writes many tokens in parallel: about 2x faster than MinerU2.5 at nearly the same accuracy. Reports ParseBench, not OmniDocBench v1.6 or olmOCR-Bench.
OmniDocBench v1.6
—
Not reported
olmOCR-Bench
—
Not reported
Parameters
2.7B
OCR specialist
Memory to run
~6.4 GB
16-bit weights plus about 1 GB
Benchmarks
No published score on these benchmarks yet. Know of one? Send it.
Specifications
- Lab
- OpenDataLab (Shanghai AI Lab) (CN)
- Type
- OCR / document parsing specialist
- Total parameters
- 2.7B
- License
- MIT
- Memory to run
- ~6.4 GB at 16-bit
- Released
- 2026-03-24
Closest alternatives
- OXR 1.0 (1.5B, 97.03 on OmniDocBench, OXR Open Source License)
- Youtu-Parsing-Omni (5.3B, 96.96 on OmniDocBench, Youtu-Parsing license)
- TeleOCR (1.4B, 96.91 on OmniDocBench, Apache-2.0)
- Xiaomi-OCR-0 (870M, 96.83 on OmniDocBench, Apache-2.0)
Sources
Something wrong here? Report it.