Community fork of llama.cpp
The goal is simple: more functionality, and the fastest llama.cpp around. And help the community with a single fast llama.cpp fork instead of many competing ones.
Halo Box keeps two forks, and which one you want depends on your hardware:
| Fork | What it is |
|---|---|
| halo-box/llama.cpp (this repo) | Stays close to mainline. Tracks upstream master and adds features and speedups on top, without diverging from how upstream works. |
| halo-box/strix-llama.cpp | Purely optimised for AMD Strix Halo machines (Ryzen AI Max+, RDNA 3.5 / gfx1151). Free to diverge from upstream wherever that buys speed. |
Use this repo if you want upstream behaviour plus extras. Use strix-llama.cpp if you run a Strix Halo box and
want every last token/s out of it.
Upstream behaviour is unchanged - this is a superset, not a rewrite. On top of it, this fork carries:
- Speculative prefill (
--spec-prefill) - a small draft model scores prompt tokens by attention importance so the target model only prefills the ones that matter, cutting time-to-first-token on long prompts. - MTP draft head for speculative decoding - use a model's own multi-token-prediction head as the draft model, instead of loading a second model alongside it.
- N-gram table on disk (
--ngram-on-disk) - keeps a model's n-gram hash-embedding table (28.8 GB on Qwen3.8-Flash-Next) off the memory budget entirely, reading only the rows each batch actually gathers. - Vulkan fixes and tuning for RDNA 3.5 - driver-gated coopmat LDS stride padding, UMA bulk readback gated on host-cached mappings, IQ3_S mat-vec at batch sizes > 4, and a radix top-k kernel.
- A
hiddenserver preset option - keep a model loadable by name while omitting it fromGET /models. LLAMA_GRAPH_TIMING=1- report where the CPU time of a decode actually goes (graph build, alloc, inputs).
Work lands on halo/* branches, and upstream is merged in regularly. Anything generally useful is sent upstream;
what stays here is either not yet ready to go up, or too niche for mainline.
LLM inference in C/C++
ggml / ops / maintainer PRs / dev stats / lib llama API / llama-server REST API
A few options to get llama.cpp installed on your machine:
- Visit https://llama.app and follow the instructions
- Run with Docker - see our Docker documentation
- Download pre-built binaries from the releases page
- Build from source by cloning this repository - check out our build guide
Once installed:
# Download and run a model directly from Hugging Face
llama cli -hf ggml-org/Qwen3.5-0.8B-GGUF
# Launch OpenAI-compatible API server
llama serve -hf ggml-org/Qwen3.5-0.8B-GGUF
VLM session with llama cli
|
Built-in web UI against llama serve
|
The main goal of llama.cpp is to enable LLM (and VLM) inference with minimal setup and state-of-the-art performance on
a wide range of hardware - locally and in the cloud.
- Plain C/C++ implementation without any dependencies
- Apple silicon is a first-class citizen - optimized via ARM NEON, Accelerate and Metal frameworks
- AVX, AVX2, AVX512 and AMX support for x86 architectures
- RVV, ZVFH, ZFH, ZICBOP and ZIHINTPAUSE support for RISC-V architectures
- 1.5-bit, 2-bit, 3-bit, 4-bit, 5-bit, 6-bit, and 8-bit integer quantization for faster inference and reduced memory use
- Custom CUDA kernels for running LLMs on NVIDIA GPUs (support for AMD GPUs via HIP and Moore Threads GPUs via MUSA)
- Vulkan and SYCL backend support
- CPU+GPU hybrid inference to partially accelerate models larger than the total VRAM capacity
The llama.cpp project is build on top of the ggml library.
| Backend | Target devices |
|---|---|
| BLAS | All |
| BLIS | All |
| CANN | Ascend NPU |
| CUDA | Nvidia GPU |
| HIP | AMD GPU |
| Hexagon [In Progress] | Snapdragon |
| IBM zDNN | IBM Z & LinuxONE |
| MUSA | Moore Threads GPU |
| Metal | Apple Silicon |
| OpenCL | Adreno GPU |
| OpenVINO [In Progress] | Intel CPUs, GPUs, and NPUs |
| RPC | All |
| SYCL | Intel GPU |
| VirtGPU | VirtGPU APIR |
| Vulkan | GPU |
| WebGPU | All |
| ZenDNN | AMD CPU |
- How to build
- Running on Docker
- Build on Android
- Multi-GPU usage
- Performance troubleshooting
- GGML tips & tricks
- XCFramework
- Completions
- Models
- Release process
- Contributors can open PRs
- Collaborators will be invited based on contributions
- Maintainers can push to branches in the
llama.cpprepo and merge PRs into themasterbranch - Any help with managing issues, PRs and projects is very appreciated!
- Read the CONTRIBUTING.md for more information
- yhirose/cpp-httplib - Single-header HTTP server, used by
llama-server- MIT license - nothings/stb - Single-header image format decoder, used by multimodal subsystem - Public domain
- nlohmann/json - Single-header JSON library, used by various tools/examples - MIT License
- mackron/miniaudio - Single-header audio format decoder, used by multimodal subsystem - Public domain
- sheredom/subprocess.h - Single-header process launching solution for C and C++ - Public domain