tinyanalysis

OCR models / Jina AI

jina-vlm

Jina AI · released 2025-12-03 · data checked 2026-11-06

jina-vlm is a 2.5B-parameter general vision-language model from Jina AI (DE), released 2025-12-03 under the CC-BY-NC-4.0 license. It has no published score on OmniDocBench v1.6 or olmOCR-Bench yet. It needs about 6 GB of memory at 16-bit.

Small multilingual vision-language model, not a document parser. Reports DocVQA 90.6 and OCRBench 778, but not OmniDocBench or olmOCR-Bench. For OCR, see jina-ocr-v1.

OmniDocBench v1.6
—
Not reported
olmOCR-Bench
—
Not reported
Parameters
2.5B
General vision-language model
Memory to run
~6 GB
16-bit weights plus about 1 GB

Benchmarks

No published score on these benchmarks yet. Know of one? Send it.

See how to read the ranking.

Specifications

Lab
Jina AI (DE)
Type
General vision-language model
Total parameters
2.5B
License
CC-BY-NC-4.0
Non-commercial use only.
Memory to run
~6 GB at 16-bit
Released
2025-12-03
Weights and usage on Hugging Face ↗

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.