llama.cpp
rank 01 in local models · last updated · get your badge · is this your tool?
numptr says9.0/10
the engine everyone else is quietly running. fastest of the lot once you've set it up right.
- pricing
- free (MIT): you run it on your own hardware
- free tier
- free and open source (MIT); no account, no paid tier
- pricing source
- https://github.com/ggml-org/llama.cpp (checked 2026-10-19)
the boring-but-important bits, all sourced. you're welcome.
full specs: llama-cpp.specs
numptr specs llama-cpp
- best for
- devs who want the fastest local inference and full control of the build
- open source
- yes (MIT)
- runs
- your machine: CLI and local server
- models
- open models (GGUF)
- official site
- https://github.com/ggml-org/llama.cpp
- affiliate / referral
- none found. and no, it doesn't touch the rank.
# checked by numptr, not the vendor's marketing team.
# verified 2026-10-19 · github.com