Grepai Embeddings Ollama

This walks you through setting up Ollama as your embedding provider for GrepAI, which means your code never leaves your machine for semantic search. The guide covers four models with real tradeoffs: nomic-embed-text is the sweet spot at 274MB and fast, while mxbai-embed-large gives you maximum accuracy at 670MB. You get actual configuration snippets, performance tuning advice like keeping models loaded to avoid the 5-minute timeout, and troubleshooting for connection issues. Honestly the most useful part is the model comparison table, it tells you exactly what RAM you need and whether to prioritize speed or quality. Good for teams that can't send code to external APIs or don't want to pay embedding costs.