MiniCPM or Gemma: which model to put on the device
Both are meant for the device itself, a laptop, a phone or a mini PC, and both work without a GPU. Gemma installs with one Ollama command, goes down to 270M and up to 31B, understands images in its larger versions, and adds CodeGemma for code and FunctionGemma 270M for function calling, but its licensing depends on the version: Apache 2.0 for Gemma 4 and DiffusionGemma, the Gemma Terms of Use for earlier releases, CodeGemma and FunctionGemma. MiniCPM is Apache 2.0 throughout apart from the first MiniCPM-2B and MiniCPM 2.0, which required registration, and MiniCPM5 at 1B and 2B is built for tool calling and long context; it is not in Ollama. MiniCPM was updated in September 2026 and Gemma in June 2026, and the catalog claims Russian support for neither.
Comparison based on catalog data
| Parameter | MiniCPM | Gemma |
|---|---|---|
| Category | Text | Text, Image + text, Code |
| Developer | OpenBMB (ModelBest and Tsinghua University), China | Google, USA |
| Releases | Feb 2024 – Sep 2026 | Feb 2024 – Jun 2026 |
| Sizes | 0.5B – 8B | 270M – 31B |
| Hardware | Laptop, 1 GPU | Laptop, 1 GPU |
| Commercial use | Commercial use allowed | Commercial use allowed |
| License | Apache 2.0 (the first MiniCPM-2B and MiniCPM 2.0 had their own MiniCPM license requiring registration for commercial use) | Gemma 4 and DiffusionGemma: Apache 2.0; earlier versions, CodeGemma and FunctionGemma: Gemma Terms of Use |
| Russian | Not supported | Not stated |
| Ollama | No | Yes |
| Without GPU | Yes | Yes |
| Tasks |
|
|
Choose MiniCPM if
- You need simple agents and tool calling on low-end hardware
- Apache 2.0 across the current versions matters to you
- Data extraction and text classification with no cloud involved
Choose Gemma if
- You want a one-command install through Ollama
- You need to read photos of documents and receipts: larger Gemma handles images
- You need a size outside the MiniCPM range: from 270M, or up to 31B
Other comparisons
- Gemma or Llama: which to choose for business
- Gemma or Phi: small models for modest hardware
- Qwen or GigaChat: which to choose for business
- Qwen or Llama: which to choose for business
- GigaChat or YandexGPT: which to choose for business
- DeepSeek or Qwen: which to choose for business
- Mistral or Llama: which to choose for business
- gpt-oss or Qwen: which one to run on your own server
- Llama or DeepSeek: which to run inside your own perimeter
- Mistral or Qwen: which to choose for business
- YandexGPT or Qwen: which to choose for Russian
- T-Pro or GigaChat: which one for a Russian company
Need a model for your task?
An open model can run on your own server: data stays in-house, there is no per-request fee, and the model can be fine-tuned on your documents.
- SelectThe model and size for your task and hardware budget
- DeployOn your server or in a closed network, with an API
- Fine-tuneOn your data, or connect a knowledge base
- IntegrateInto your CRM, ERP, bot, website or team chat