Skip to content

Run MiniCPM-V 4.7, a Free 35B AI, on Your Laptop or iPhone

A free, open-source 35B model with only about 3B parameters active at a time: how to run it with Ollama or build OpenBMB's iPhone app, what OpenBMB's own benchmarks say, and the catch

Alex Chen@nocodealex · 45,000+ follow on Instagram

3min read

2prompts to copy

10numbered steps

6sections

MiniCPM-V 4.7 is the newest model from OpenBMB, released on 12 October 2026. It is the first mixture-of-experts model in the MiniCPM-V series: 35 billion parameters in total, but only about 3 billion of them wake up for each token (that is the A3B in 35B-A3B). It reads single images, several images and video, and its licence is Apache 2.0, so it is free and open source.

Run it on your laptop

Install Ollama from ollama.com. It is free.
Open a terminal and type: ollama run openbmb/minicpm-v4.7
The first run downloads about 22 GB, the 35B-A3B default. Check your free disk space before you press Enter.
Short on space? Type this instead for the small 1B model, about 1.8 GB: ollama run openbmb/minicpm-v4.7:1b
When the prompt appears, chat with it in the terminal.

OR PASTE THIS INTO CLAUDE CODE

Check how much free disk space I have. Then run `ollama run openbmb/minicpm-v4.7:1b` (the small 1.8 GB version) and tell me when the chat is ready. Ask me before downloading the 22 GB 35B version.

Keep reading

Free, for an email.

Unlocks 5 more sections, 1 copy-paste prompt, the document version, and every other guide on the site, for free. Enter your email once to keep reading.

45,000+ follow where these guides come from.No spam. Unsubscribe in one click.

Read next

Run a 70B AI Model on a 4GB Graphics Card: Set Up the Free AirLLM Library

Big open AI models need an expensive graphics card, so most people never try one. AirLLM is a free, open-source Python library that runs a 70 billion parameter model on a single 4GB GPU by keeping only one layer of it in memory at a time. The install command, the code to paste in, ready-made prompts for Claude Code, and the model sizes from its own README

Open Design, the Free Open-Source Claude Design: Websites, App Screens and Decks Without Burning Your Claude Limit

Claude Design draws from the same usage limit as Claude Code, and a full deck uses more of it than a typical message. Open Design is a free, open-source alternative with 99,000 GitHub stars that runs on the model you choose, DeepSeek included. What it makes, the 151 brand styles it ships with, and how to set it up.

FluidVoice, the Free Open-Source Wispr Flow: Ramble at Claude, Get a Clean Prompt, Even Offline

Press one key, talk, and a cleaned-up version of what you said is typed into any app on your Mac, from Claude to Slack. FluidVoice does it for free, with the speech model running on your own Mac, so it works with no internet. Why it can replace a Wispr Flow subscription, how to set it up, and how to use it with Claude.