What MiniCPM-V 4.7 is
OpenBMB published MiniCPM-V 4.7 on Hugging Face in October 2026. According to the model page:
- It is the first mixture-of-experts model in the MiniCPM-V series: 35 billion parameters in total.
- It reads one or several images, video, text in photos and PDFs.
- It runs on iOS, Android and HarmonyOS. On an iPhone 17 Pro Max, OpenBMB shows more than 13 tokens per second with 3 GB of memory.
Who it is for
- Field teams: read an invoice, a report or a document with the camera, without depending on the network.
- Records and archives: take the text from PDFs and images into the system.
Why through Rota Nacional
- We bring the model to your organization's app or run it on Rota's workers, on the same API.
- By default, CPF and CNPJ numbers, email addresses, phone numbers and names leave the text before the model, when it runs through the API.
How it works
You describe the case
The documents, the devices and the volume.
We test with your samples
OpenBMB does not list the model's languages. So we measure reading in Brazilian Portuguese with your own documents.
We connect it to your access
The model joins your app or the API you already use.
Good to know
- Apache-2.0 license, checked on Hugging Face on October 17, 2026.
- The figures come from OpenBMB, with no third-party evaluation. The page does not say which build used the 3 GB; the smallest GGUF file of the model has 14.5 GB.
- On-demand deployment: the model goes live when an organization asks for it.