Rimo
Open models. Your PC. No cloud.
Rimo finds language models on Hugging Face, tells you which version fits your graphics card, downloads it and lets you chat with it, entirely on your own computer.
Thousands of open AI models are published on Hugging Face, but running one yourself usually means picking files, quantisations and settings by hand, and guessing whether it will even fit on your hardware.
Rimo does that for you. Search, pick the recommended version, click download, start chatting. Your conversations never leave your PC.
What Rimo does
01
Search all of Hugging Face
Every repository with GGUF files, sorted by popularity, trend, likes or date. Or paste a link and go straight to the model.
02
Knows your hardware
For every version Rimo estimates the memory it needs and shows whether it runs fully on your GPU, partly on the CPU, or not at all.
03
One-click downloads
Downloads run in the background with progress, speed and remaining time, and can be cancelled at any moment.
04
Chat, fully local
Answers stream in live with formatting and code blocks. All chats are saved on your PC and can be renamed or deleted.
05
Watch it think
Reasoning models such as Qwen3 show their thought process in a collapsible block, and thinking can be switched off per model.
06
Settings per model
System prompt, temperature and context length for each model, with a live estimate of how much graphics memory the context needs.
VRAM check
The right version for your graphics card.
Most models come in a dozen versions, from heavily compressed to nearly lossless. Bigger is better, until it no longer fits.
Rimo reads your graphics card, adds up the model, the memory for the conversation and the runtime overhead, and recommends the best quality that still runs fully on your GPU.
Questions
What do I need?
Windows and Ollama, the free local model runtime. Rimo uses Ollama to download and run the models. An NVIDIA graphics card is recommended; without one, models run on the CPU and are slower.
Does Rimo really run every model from Hugging Face?
Every model that is published as GGUF, which covers almost all well-known chat models such as Llama, Qwen, Mistral, Gemma and DeepSeek. Models that only exist as raw weights, image or audio models, and architectures Ollama does not support yet will not work.
Does anything leave my computer?
Only the search on Hugging Face and the model download. Your chats, settings and models stay on your PC.
What does it cost?
Nothing. Rimo is free and its source code is on GitHub.
Try Rimo
Source code, build instructions and updates are on GitHub.