Library / SDK
Inference & Serving
GGML
1Low 2level 3C 4tensor 5inference 6library
the six Ws · specification
W1
Who
Created by Georgi Gerganov and maintained by the ggml-org community.
W2
What
GGML is a low level C tensor library for machine learning inference with built in quantization support.
W3
Where
Hosted on GitHub at ggml-org/ggml, migrated from the original ggerganov/ggml repository.
W4
When
Released in 2022 and now underpins llama.cpp and whisper.cpp.
W5
Why
It provides the lightweight, dependency free tensor and quantization foundation that made running large models on ordinary laptops practical.
W6
With
Has no external dependencies and offers optional CUDA, Metal and Vulkan backends.
W7
Watch
MIT licensed, open source, maintained by Georgi Gerganov and the ggml-org community, roughly 15k GitHub stars.
alternatives
works with