Runs large Mixture-of-Experts models such as GLM, DeepSeek V4, and Kimi on consumer hardware by streaming routed experts from SSD, with an OpenAI-compatible local API.
Use it now
Keep it with up to 9 other tools you're setting up, then share them all as one link.
Try this first
Ask your coding agent:
- Multi-SSD Mirrors
Use Colibri to stream model copies: ‹what you need›
Install the CLI
Install it as its own instructions say.
Its page: github.com/JustVugg/colibri
Call the API
Auth None needed
Its page: github.com/JustVugg/colibri
Compare
Press Compare to add it to the tray at the bottom of the window, then pick up to 2 others anywhere on the site.
- Access
- CLIAPI
- Auth
- None
- Skills
- 4
Skills
Expert Streaming
Keeps dense weights resident in RAM and streams routed experts from disk on demand, with LRU and pinned hot-store caches.
OpenAI-Compatible API
Serves chat and completion endpoints with SSE streaming and tool calls through coli serve or coli web.
Local Cluster Mode
Spreads routed expert execution across worker machines via a coordinator, keeping routing and KV state local.
Multi-SSD Mirrors
Streams model copies from more than one drive, validating mirrors at startup and falling back to the primary on errors.