MiniCPM

Health v2 · maintenance only

Adoption not part of the score

11232 stars · 771 forks observed · 2026-09-24

What it is AI-extracted, prompt v1, taxonomy v1, 2026-10-12, confidence 0.97

MiniCPM is a family of compact multimodal large language models designed for efficient image, video, and audio understanding on mobile devices. It provides both vision-focused (MiniCPM-V) and omnidirectional (MiniCPM-o) variants with open-source deployment code.

Use cases

• perform image captioning on a smartphone
• analyze video content directly on a mobile device
• run multimodal question answering with both text and visual inputs
• enable real-time video and audio interaction in mobile apps
• extract information from screenshots or photos without cloud services

When to choose

• You need on-device multimodal AI with low latency and minimal resource usage.
• Your application runs on Android, iOS, or other mobile platforms and cannot rely on server-side inference.
• You prefer an open-source model with a small footprint (≈1‑9B parameters).

When to avoid

• You require the highest possible accuracy regardless of model size.
• Your deployment environment is limited to pure CPU without any hardware acceleration.
• You need extensive support for languages or modalities not covered by MiniCPM.

Facets

library · maturity active

llm-inference audio-processing video-processing nlp artificial-intelligence large-language-models computer-vision windows python pocket-sized efficient multimodal-llm image-video-understanding on-device mini-cpm vision-language multimodal video audio android ios linux mobile

Member repositories

For agents

markdown · JSON · MCP: product_card(name="OpenBMB/MiniCPM")