LocalAI can be built as a container image or as a single, portable binary. Note that the some model architectures might require Python libraries, which are not included in the binary. The binary contains only the core backends written in Go and C++.

LocalAI’s extensible architecture allows you to add your own backends, which can be written in any language, and as such the container images contains also the Python dependencies to run all the available backends (for example, in order to run backends like Diffusers that allows to generate images and videos from text).

In some cases you might want to re-build LocalAI from source (for instance to leverage Apple Silicon acceleration), or to build a custom container image with your own backends. This section contains instructions on how to build LocalAI from source.

Build LocalAI locally


In order to build LocalAI locally, you need the following requirements:

  • Golang >= 1.21
  • Cmake/make
  • GCC
  • GRPC

To install the dependencies follow the instructions below:


To build LocalAI with make:

  git clone
cd LocalAI
make build

This should produce the binary local-ai

Here is the list of the variables available that can be used to customize the build:

BUILD_TYPENoneBuild type. Available: cublas, openblas, clblas, metal,hipblas, sycl_f16, sycl_f32
GO_TAGStts stablediffusionGo tags. Available: stablediffusion, tts, tinydream
CLBLAST_DIRSpecify a CLBlast directory
CUDA_LIBPATHSpecify a CUDA library path
BUILD_API_ONLYfalseSet to true to build only the API (no backends will be built)

Container image


  • Docker or podman, or a container engine

In order to build the LocalAI container image locally you can use docker, for example:

  # build the image
docker build -t localai .
docker run localai

There are some build arguments that can be used to customize the build:

IMAGE_TYPEextrasBuild type. Available: core, extras

Example: Build on mac

Building on Mac (M1, M2 or M3) works, but you may need to install some prerequisites using brew.

The below has been tested by one mac user and found to work. Note that this doesn’t use Docker to run the server:

Install xcode from the Apps Store (needed for metalkit)

  # install build dependencies
brew install abseil cmake go grpc protobuf wget protoc-gen-go protoc-gen-go-grpc

# clone the repo
git clone

cd LocalAI

# build the binary
make build

# Download phi-2 to models/
wget -O models/phi-2.Q2_K

# Use a template from the examples
cp -rf prompt-templates/ggml-gpt4all-j.tmpl models/phi-2.Q2_K.tmpl

# Run LocalAI
./local-ai --models-path=./models/ --debug=true

# Now API is accessible at localhost:8080
curl http://localhost:8080/v1/models

curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
     "model": "phi-2.Q2_K",
     "messages": [{"role": "user", "content": "How are you?"}],
     "temperature": 0.9 

Troubleshooting mac

  • If you encounter errors regarding a missing utility metal, install Xcode from the App Store.

  • After the installation of Xcode, if you receive a xcrun error 'xcrun: error: unable to find utility "metal", not a developer tool or in PATH'. You might have installed the Xcode command line tools before installing Xcode, the former one is pointing to an incomplete SDK.

  # print /Library/Developer/CommandLineTools, if command line tools were installed in advance
xcode-select --print-path

# point to a complete SDK
sudo xcode-select --switch /Applications/
  • If completions are slow, ensure that gpu-layers in your model yaml matches the number of layers from the model in use (or simply use a high number such as 256).

  • If you a get a compile error: error: only virtual member functions can be marked 'final', reinstall all the necessary brew packages, clean the build, and try again.

  # reinstall build dependencies
brew reinstall abseil cmake go grpc protobuf wget

make clean

make build

Requirements: OpenCV, Gomp

Image generation requires GO_TAGS=stablediffusion or GO_TAGS=tinydream to be set during build:

  make GO_TAGS=stablediffusion build

Build with Text to audio support

Requirements: piper-phonemize

Text to audio support is experimental and requires GO_TAGS=tts to be set during build:

  make GO_TAGS=tts build



Software acceleration.

Requirements: OpenBLAS

  make BUILD_TYPE=openblas build


Nvidia Acceleration.

Requirement: Nvidia CUDA toolkit

Note: CuBLAS support is experimental, and has not been tested on real HW. please report any issues you find!

  make BUILD_TYPE=cublas build

More informations available in the upstream PR:

Hipblas (AMD GPU with ROCm on Arch Linux)


  pacman -S base-devel git rocm-hip-sdk rocm-opencl-sdk opencv clblast grpc

Library links:

  export CGO_CFLAGS="-I/usr/include/opencv4"
export CGO_CXXFLAGS="-I/usr/include/opencv4"
export CGO_LDFLAGS="-L/opt/rocm/hip/lib -lamdhip64 -L/opt/rocm/lib -lOpenCL -L/usr/lib -lclblast -lrocblas -lhipblas -lrocrand -lomp -O3 --rtlib=compiler-rt -unwindlib=libgcc -lhipblas -lrocblas --hip-link"


  make BUILD_TYPE=hipblas GPU_TARGETS=gfx1030


AMD/Intel GPU acceleration.

Requirement: OpenCL, CLBlast

  make BUILD_TYPE=clblas build

To specify a clblast dir set: CLBLAST_DIR

Intel GPU acceleration

Intel GPU acceleration is supported via SYCL.

Requirements: Intel oneAPI Base Toolkit (see also llama.cpp setup installations instructions)

  make BUILD_TYPE=sycl_f16 build # for float16
make BUILD_TYPE=sycl_f32 build # for float32

Metal (Apple Silicon)

  make build

# correct build type is automatically used on mac (BUILD_TYPE=metal)
# Set `gpu_layers: 256` (or equal to the number of model layers) to your YAML model config file and `f16: true`

Windows compatibility

Make sure to give enough resources to the running container. See


More advanced build options are available, for instance to build only a single backend.

Build only a single backend

You can control the backends that are built by setting the GRPC_BACKENDS environment variable. For instance, to build only the llama-cpp backend only:

  make GRPC_BACKENDS=backend-assets/grpc/llama-cpp build

By default, all the backends are built.

Specific llama.cpp version

To build with a specific version of llama.cpp, set CPPLLAMA_VERSION to the tag or wanted sha:

  CPPLLAMA_VERSION=<sha> make build

Last updated 01 Jun 2024, 18:59 +0200 . history