#3 · Primary category： Foundation Models
MiniCPM-V
A Pocket-Sized MLLM for Ultra-Efficient Image and Video Understanding on Your Phone
Project last updated：08/26/26
GitHub Stars
26.3K
Forks
2.1K
Contributors
38
License
Apache-2.0
Why we included this project
MiniCPM-V is a family of multimodal LLMs that trades raw size for practicality: the models are small enough to run on phones, tablets, and edge hardware while still handling image, video, and text inputs. The 1.3B-parameter MiniCPM-V 4.6 compresses visual tokens early in the pipeline, cutting encoding cost by more than half without sacrificing throughput, and the repo includes edge adaptation code for iOS, Android, and HarmonyOS. MiniCPM-o 4.5 pushes the same family further, adding real-time interaction with streaming video and audio plus speech output, so a model can see, listen, and speak in a live conversation. Since the models plug into Ollama, llama.cpp, and vLLM, you can prototype locally and later serve them without changing your stack.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
transformers
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
CLIP
CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image
generative-models
Generative Models by Stability AI
unilm
Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
DeepSeek-Coder
DeepSeek Coder: Let the Code Write Itself