llama.cpp
#7 in Lokale LLM-RuntimesUnknown · siet 10. März 2023 · 2× · tolest 14. Okt. 2026
38
Momentum
llama.cpp is an open-source C/C++ library for large language model inference, especially in the GGUF format, running on CPUs and GPUs without Python dependencies. The project was first released in March 2023 by Georgi Gerganov and is developed today in the ggml-org repository on GitHub. It includes the llama-server HTTP server with an OpenAI-compatible API.
Momentum-Verloop
16.07.14.10.
Features
| License | MIT |
| Platform | CPU, Apple Silicon (Metal), CUDA, HIP, Vulkan, SYCL, CANN, etc. |
| Price | Free (open source, MIT license) |
| Protocol Compatibility | OpenAI-compatible API (llama-server), plus Anthropic Messages API (/v1/messages) |
| Release Date | March 10, 2023 |
| Supported Models/Providers | Dozens of model families incl. LLaMA, Mistral, Qwen, Gemma; GGUF format |