llama-coder

Why we included this project

Llama Coder is a VS Code extension that runs code completion entirely on your own hardware through Ollama, using open models like CodeLlama and Stable Code. The README is refreshingly honest about what it takes to run well: each model is listed with its RAM or VRAM footprint, and the author recommends Apple Silicon or an RTX 4090 for the smoothest results. Setup covers both a simple local install and a remote configuration where one GPU machine serves completions to other workstations. That makes it a practical pick for developers and small teams who want Copilot-like autocomplete without sending code to a cloud service, and without telemetry. It's a working extension, not a proof of concept, so expect to install it, point it at Ollama, and choose a model size that.