Run Local Models Alongside Cloud AIs

Cloud models are powerful, but sometimes you need data to stay local. Maybe it's a sensitive project. Maybe you want to experiment without API costs. Maybe you just like the idea of owning the entire stack. LobeHub v0.127.0 now supports Ollama, giving you the same chat experience whether your model lives in the cloud or on your machine.

No separate interface to learn. No workflow fragmentation. Just point LobeHub at your local Ollama instance and start chatting.

Connect Your Local Models in One Line

Getting started is straightforward. If you already have Ollama running, connect LobeHub with a single Docker command:

bash

docker run -d -p 3210:3210 -e OLLAMA_PROXY_URL=http://host.docker.internal:11434/v1 lobehub/lobe-chat

That's it. LobeHub detects your local models and makes them available in the same model switcher you use for GPT-4, Claude, and others. Mix cloud and local models in the same workspace depending on what each conversation needs.

When to Use Local Models

Privacy-first work: Keep sensitive conversations on your machine
Cost control: No per-token charges for experimentation
Offline access: Continue working without internet connectivity
Model testing: Evaluate open-source models before production deployment

Improvements and fixes

Added automatic model discovery from Ollama endpoints
Fixed streaming response handling for local model compatibility
Improved error handling when Ollama service is unreachable

Credits

Huge thanks to the community contributor who made Ollama integration possible, and to the Ollama team for building accessible local AI infrastructure.