The llama.cpp and ggml libraries have reached version 1.0.0. The ggml-org team says future API and ABI incompatibilities in both projects will trigger major version bumps. The change gives developers who use the libraries a clearer way to track compatibility.
llama.cpp runs multimodal large language models across hardware ranging from consumer machines to enterprise servers. ggml is its underlying machine learning engine and tensor library. The team says many downstream applications have come to depend on its work during the project’s 3.5 years of development.
The team’s versioning rules cover breaking changes to the public C API, backward-compatible additions and updates, and bug fixes that leave the API unchanged. It says it will be stricter about making compatibility boundaries visible to developers who use the libraries.
Read nextMeituan releases LongCat-2.5 weights after training on Chinese chipsModel support spans more than 150 variants
According to the team, llama.cpp supports more than 150 model architecture variants and all mainstream speculative decoding methods. ggml uses dedicated backends for different hardware. Each backend can be compiled as a dynamic library and optionally loaded at runtime.
Version 1.0.0 is released for both libraries. The team also lists pre-release artifacts for Linux, macOS, Windows, Android and iOS, along with distribution through Homebrew, conda-forge, MacPorts and an XCFramework for iOS and macOS.



