What is llama.vim?

llama.vim is a plugin for the Vim and Neovim text editors that adds AI-assisted code and text completion, running entirely on your own computer without sending data to any external service. It works by connecting to a locally running AI model server called llama.cpp, which runs open-source language models on your own hardware. When you are typing in Vim, the plugin watches your cursor position and automatically suggests completions inline. You can accept a full suggestion by pressing Tab, or accept just the first line with Shift+Tab. There is also an instruction-based editing mode where you can describe a change in plain text and have the AI apply it to a selected block of code. The plugin is designed to work on consumer-grade hardware, not just high-end machines. It adjusts its behavior based on how much graphics memory your computer has, and the README provides specific recommended settings for machines with 8GB, 16GB, or 64GB of video memory. It also includes a technique for reusing context across files, which helps keep responses fast even in large projects. Installing it requires two steps: adding the plugin to Vim using a standard plugin manager like vim-plug or lazy.nvim, and separately installing and running the llama.cpp server, which is available through package managers on Mac and Windows or as a downloadable binary for other systems. The plugin is maintained by the ggml-org organization, the same group behind llama.cpp. A companion extension for VS Code is also available separately.