OCR models / Alibaba (Ovis)
Ovis2.6 30B-A3B
Alibaba (Ovis) · released 2026-02-12 · data checked 2026-11-06
Ovis2.6 30B-A3B is a 31.4B-parameter general vision-language model from Alibaba (Ovis) (CN), released 2026-02-12 under the Apache-2.0 license. It scores 93.70 on OmniDocBench v1.6 (#17 of 33). It needs about 63.8 GB of memory at 16-bit.
A general vision-language model, not an OCR specialist; listed for comparison.
OmniDocBench v1.6
93.70
#17 of 33 models
olmOCR-Bench
—
Not reported
Parameters
31.4B
3B active · Mixture-of-Experts
Memory to run
~63.8 GB
16-bit weights plus about 1 GB
Specifications
- Lab
- Alibaba (Ovis) (CN)
- Type
- General vision-language model
- Total parameters
- 31.4B
- Active parameters
- 3B
- License
- Apache-2.0
- Memory to run
- ~63.8 GB at 16-bit
- Released
- 2026-02-12
Closest alternatives
- Youtu-Parsing (2.5B, 93.74 on OmniDocBench, Youtu-Parsing license)
- Qianfan-OCR (4.7B, 93.90 on OmniDocBench, Apache-2.0)
- Infinity-Parser2 Pro (35.1B, 93.95 on OmniDocBench, Apache-2.0)
- Unlimited-OCR (3.3B, 94.00 on OmniDocBench, MIT)
Sources
Something wrong here? Report it.