OCR models / OpenBMB
MiniCPM-V 4.7 35B-A3B
OpenBMB · released 2026-10-07 · data checked 2026-11-06
MiniCPM-V 4.7 35B-A3B is a 35.2B-parameter general vision-language model from OpenBMB (CN), released 2026-10-07 under the Apache-2.0 license. It scores 94.24 on OmniDocBench v1.6 (#12 of 33). It needs about 71.4 GB of memory at 16-bit.
A general vision-language model (Mixture-of-Experts, about 3B active) that scores close to OCR specialists. In the same chart OpenBMB measures Qwen3.6 35B-A3B at 80.17 and Gemma 4 31B at 79.17.
OmniDocBench v1.6
94.24
#12 of 33 models
olmOCR-Bench
—
Not reported
Parameters
35.2B
3B active · Mixture-of-Experts
Memory to run
~71.4 GB
16-bit weights plus about 1 GB
Benchmarks
OmniDocBench v1.6 Lab-reported94.24
Read from the chart image on the model card (4x visual token compression). Best: OXR 1.0 97.03
The tick marks the best score on each benchmark. See how to read the ranking.
Specifications
- Lab
- OpenBMB (CN)
- Type
- General vision-language model
- Total parameters
- 35.2B
- Active parameters
- 3B
- License
- Apache-2.0
- Memory to run
- ~71.4 GB at 16-bit
- Released
- 2026-10-07
Closest alternatives
- Unlimited-OCR (3.3B, 94.00 on OmniDocBench, MIT)
- Infinity-Parser2 Pro (35.1B, 93.95 on OmniDocBench, Apache-2.0)
- Qianfan-OCR (4.7B, 93.90 on OmniDocBench, Apache-2.0)
- HunyuanOCR 1.5 (1.1B, 94.74 on OmniDocBench, Tencent Hunyuan Community License)
Sources
- https://huggingface.co/openbmb/MiniCPM-V-4.7-35B-A3B
- https://raw.githubusercontent.com/OpenBMB/MiniCPM-o/main/assets/minicpmv4.7/overall_performance_35ba3b_instruct.png
Something wrong here? Report it.