Find the best local LLM for your hardware, ranked by benchmarks

This project does one thing: takes your hardware profile and returns a ranked list of local LLMs that will actually fit and run well on that machine. The differentiator over just reading the GGUF subreddit or squinting at leaderboard tables is that it combines benchmark scores with real hardware constraints — VRAM, RAM, and compute class — so the recommendation is actionable, not aspirational. For a solo founder or indie hacker evaluating which model to bundle into an offline product, this replaces a multi-hour research detour. The repo is early-stage, so expect rough edges: the hardware input interface is minimal, model coverage is not exhaustive yet, and benchmark freshness depends on manual updates. But the core loop is useful enough to bookmark right now if local inference is on your roadmap. Reservation: community-contributed benchmark data means quality varies. Verify any recommendation against a second source before committing to a model architecture. -> Best for: indie hacker or AI engineer building local-first or offline-capable products