ReadingResearchRadarInvestment framework
Sign in / Sign up
← Back to reading
NewsMaterial1 min readDisconfirmAI Editorial

Open-source LLM inference engine llama.cpp releases v1.0.0, establishing API compatibility rules

After three and a half years, llama.cpp reaches its first major version, defining strict API stability guarantees to reduce integration risks for developers.

2026-11-16 13:43 GMT+8

llama.cpp, the C/C++ based open-source Large Language Model (LLM) inference engine, released version 1.0.0 on November 13, 2026. This marks the project's first major version number update in approximately three and a half years, signaling a transition from rapid iteration to stable maintenance. The underlying tensor computation library, ggml, was simultaneously updated to v1.0.0.

The development team stated that as llama.cpp sees widespread adoption across various applications, providing flexible and stable APIs has become crucial. To address this, the project established strict semantic versioning rules: major version numbers are incremented only when breaking changes occur to public APIs; backward-compatible additions such as new features, backend updates, or model support fall under minor versions; and bug fixes without API changes are released as patches. This policy provides clear stability expectations for developers relying on the engine for local deployment and integration.

Technically, v1.0.0 removed previously deprecated APIs and introduced optimizations for FP8 KV cache and Mixture-of-Experts (MoE) cache improvements. Additionally, the new version added support for several AI models, including Kimi-K3.1-Pro, Ripple-65B-Preview, xing4_0, and drama-3 TTS.

Sources

  • pc.watch.impress.co.jpPC Watch↗
Sign in / Sign up
ReadingResearchRadarInvestment framework
Continue researching →