{"id":1238,"date":"2026-09-17T16:56:51","date_gmt":"2026-09-17T07:56:51","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/engine-llama-cpp-en\/"},"modified":"2026-10-29T01:40:09","modified_gmt":"2026-10-28T16:40:09","slug":"engine-llama-cpp-en","status":"publish","type":"page","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-llama-cpp-en\/","title":{"rendered":"llama.cpp: Releases and Overview"},"content":{"rendered":"

LLM inference engine written in C\/C++. Runs GGUF models on CPU and GPU (CUDA \/ Metal \/ Vulkan \/ ROCm) and underpins much of the local-LLM ecosystem, including Ollama, LM Studio and KoboldCpp.<\/p>\n

Category: Inference engines and runtimes. Part of our engine and tool tracker<\/a>.<\/em><\/p>\n

At a Glance<\/h2>\n
\n\n\n\n\n\n\n\n\n\n\n
Item<\/th>\nValue<\/th>\n<\/tr>\n<\/thead>\n
Repository<\/td>\nggml-org\/llama.cpp<\/a><\/td>\n<\/tr>\n
License<\/td>\nMIT<\/td>\n<\/tr>\n
Main language<\/td>\nC++<\/td>\n<\/tr>\n
GitHub stars<\/td>\n128,500 (as of 2026-09-17)<\/td>\n<\/tr>\n
Latest release<\/td>\nv0.8.0<\/a> (2026-10-28)<\/td>\n<\/tr>\n
Install \/ docs<\/td>\nofficial documentation<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

The project describes itself as: \u201cLLM inference in C\/C++\u201d<\/p>\n

Release History<\/h2>\n

Compiled by Local Model Watch from the project’s GitHub releases. Pre-releases are omitted. Where we wrote an article about a release, it is linked in the last column; smaller releases are tracked here only.<\/em><\/p>\n

\n\n\n\n\n\n\n\n\n\n\n\n
Version<\/th>\nReleased<\/th>\nRelease notes<\/th>\nOur article<\/th>\n<\/tr>\n<\/thead>\n
v0.8.0<\/td>\n2026-10-28<\/td>\nGitHub<\/a><\/td>\n\u2014<\/td>\n<\/tr>\n
v0.7.0<\/td>\n2026-10-16<\/td>\nGitHub<\/a><\/td>\n\u2014<\/td>\n<\/tr>\n
v0.6.0<\/td>\n2026-10-06<\/td>\nGitHub<\/a><\/td>\nllama.cpp v0.6.0 Released with llama_batch_ext and Metal Boosts<\/a><\/td>\n<\/tr>\n
v0.5.0<\/td>\n2026-09-24<\/td>\nGitHub<\/a><\/td>\nllama.cpp v0.5.0 Released with Backend and Server Upgrades<\/a><\/td>\n<\/tr>\n
v0.4.1<\/td>\n2026-09-15<\/td>\nGitHub<\/a><\/td>\nllama.cpp v0.4.1 Released with Breaking Changes<\/a><\/td>\n<\/tr>\n
v0.4.0<\/td>\n2026-09-05<\/td>\nGitHub<\/a><\/td>\n\u2014<\/td>\n<\/tr>\n
v0.3.0<\/td>\n2026-08-25<\/td>\nGitHub<\/a><\/td>\n\u2014<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Speed by Version on Our CPU<\/h2>\n

Measured by Local Model Watch on our CPU-only server (Neoverse-N1, 3 threads) with llama-bench (512-token prompt, 128-token generation), always on the same model: Qwen3-0.6B-Q4_K_M.gguf<\/code> from unsloth\/Qwen3-0.6B-GGUF<\/a> (revision 50968a4468<\/code>). Versions like v0.5.0 ship no binaries, so we used the first official ubuntu-arm64 build that contains the version; \u201c+N\u201d is how many commits that build is ahead of it. These are CPU figures, not a guide to GPU speed, and small differences can be run-to-run noise (\u00b1 is the standard deviation).<\/em><\/p>\n

\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n\n\n\n
Version<\/th>\nReleased<\/th>\nBuild measured<\/th>\nPrompt<\/th>\nGeneration<\/th>\nMeasured<\/th>\n<\/tr>\n<\/thead>\n
v0.8.0<\/td>\n2026-10-28<\/td>\nb11960<\/td>\n127.9 \u00b1 0.1 tok\/s<\/td>\n42.4 \u00b1 0.4 tok\/s<\/td>\n2026-10-29<\/td>\n<\/tr>\n
v0.7.0<\/td>\n2026-10-16<\/td>\nb11669<\/td>\n128.4 \u00b1 0.1 tok\/s<\/td>\n43.1 \u00b1 0.2 tok\/s<\/td>\n2026-10-17<\/td>\n<\/tr>\n
v0.6.0<\/td>\n2026-10-06<\/td>\nb11429<\/td>\n128.4 \u00b1 0.6 tok\/s<\/td>\n43.0 \u00b1 1.4 tok\/s<\/td>\n2026-10-07<\/td>\n<\/tr>\n
v0.5.0<\/td>\n2026-09-24<\/td>\nb11146<\/td>\n129.9 \u00b1 0.4 tok\/s<\/td>\n43.4 \u00b1 0.5 tok\/s<\/td>\n2026-10-03<\/td>\n<\/tr>\n
v0.4.1<\/td>\n2026-09-15<\/td>\nb10964<\/td>\n128.5 \u00b1 1.1 tok\/s<\/td>\n43.4 \u00b1 0.3 tok\/s<\/td>\n2026-10-03<\/td>\n<\/tr>\n
v0.4.0<\/td>\n2026-09-05<\/td>\nb10809<\/td>\n129.9 \u00b1 0.2 tok\/s<\/td>\n42.4 \u00b1 0.7 tok\/s<\/td>\n2026-10-02<\/td>\n<\/tr>\n
v0.3.0<\/td>\n2026-08-25<\/td>\nb10621<\/td>\n127.5 \u00b1 0.9 tok\/s<\/td>\n43.2 \u00b1 0.1 tok\/s<\/td>\n2026-10-02<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Articles on Local Model Watch<\/h2>\n