Install this package:
emerge -a sci-ml/lemonade| Version | EAPI | Keywords | Slot |
|---|---|---|---|
| 9999 | 8 | 0 | |
| 11.5.1 | 8 | ~amd64 | 0 |
| 11.0.0-r1 | 8 | ~amd64 | 0 |
| 10.10.0-r3 | 8 | ~amd64 | 0 |
<pkgmetadata>
<maintainer type="person">
<email>iohann.s.titov@gmail.com</email>
<name>Ivan S. Titov</name>
</maintainer>
<longdescription lang="en">
Lemonade is a local AI server that exposes optimized LLMs through
OpenAI / Anthropic / Ollama compatible APIs, running inference on
AMD NPU and GPU. The C++ server core (lemond, lemonade-server) is
packaged here without the Tauri desktop wrapper; the bundled web
frontend is available behind USE=webui, and the Tauri wrapper stays
CMake-toggleable and can be added behind a USE flag later if needed.
Pairs with sci-ml/fastflowlm to drive the AMD Ryzen AI XDNA2 NPU
backend.
</longdescription>
<use>
<flag name="fastflowlm">Pull in <pkg>sci-ml/fastflowlm</pkg> as the NPU backend (renamed to system-fastflowlm from 11.0.0; kept for 10.10.0-r3)</flag>
<flag name="openrc">Install OpenRC init.d/conf.d files for a supervise-daemon-managed service</flag>
<flag name="system-fastflowlm">Reuse the system <pkg>sci-ml/fastflowlm</pkg> (/usr/bin/flm) for the FastFlowLM NPU backend instead of letting lemond auto-download the flm runtime at first NPU use</flag>
<flag name="system-kokoro">Reuse the system <pkg>sci-ml/kokoros</pkg> (/usr/bin/koko) for the Kokoro TTS backend instead of fetching an upstream prebuilt binary at runtime</flag>
<flag name="system-llamacpp">Reuse the system <pkg>sci-misc/llama-cpp</pkg> (/usr/bin/llama-server) for the llama.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes)</flag>
<flag name="system-rocm">Point lemond's ROCm runtime at the system ROCm in /usr (via ROCM_PATH) so its rocm-stable image/audio backends reuse it instead of downloading AMD's multi-GB TheRock runtime; requires a system ROCm (pulls <pkg>dev-util/hip</pkg>). Mutually exclusive with system-therock</flag>
<flag name="system-sdcpp">Reuse the system <pkg>sci-misc/stable-diffusion-cpp</pkg> (/usr/bin/sd-server) for the stable-diffusion.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes)</flag>
<flag name="system-therock">Point lemond's ROCm runtime at <pkg>dev-util/therock-bin</pkg> in /opt/therock-bin (via ROCM_PATH) instead of downloading AMD's TheRock runtime; pinned to the 7.13 line whose version matches lemond's expected ROCm. Mutually exclusive with system-rocm</flag>
<flag name="system-whispercpp">Reuse the system <pkg>app-accessibility/whisper-cpp</pkg> (/usr/bin/whisper-server) for the whisper.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes)</flag>
<flag name="systemd">Keep upstream's systemd units (lemond.service, system + user), running as a dedicated <pkg>acct-user/lemonade</pkg></flag>
<flag name="tauri">Build the Tauri desktop app (lemonade-app) -- a native webkit2gtk window wrapping the same renderer as USE=webui. Pulls a Rust toolchain (<pkg>dev-lang/rust</pkg>) plus <pkg>net-libs/webkit-gtk</pkg>, and fetches Rust crates over the network at build time</flag>
<flag name="webui">Build the bundled React web frontend (served by lemond at the server root) via <pkg>net-libs/nodejs</pkg>; runs npm/webpack over the network at build time. Without it lemond serves a static placeholder page</flag>
</use>
<upstream>
<doc>https://lemonade-server.ai/docs/</doc>
<bugs-to>https://github.com/lemonade-sdk/lemonade/issues</bugs-to>
<remote-id type="github">lemonade-sdk/lemonade</remote-id>
</upstream>
</pkgmetadata>
Manage flags for this package:
euse -i <flag> -p sci-ml/lemonade |
euse -E <flag> -p sci-ml/lemonade |
euse -D <flag> -p sci-ml/lemonade
| Flag | Description | 9999 | 11.5.1 | 11.0.0-r1 | 10.10.0-r3 |
|---|---|---|---|---|---|
| fastflowlm | Pull in <pkg>sci-ml/fastflowlm</pkg> as the NPU backend (renamed to system-fastflowlm from 11.0.0; kept for 10.10.0-r3) | ✗ | ✗ | ✗ | ✓ |
| openrc | Install OpenRC init.d/conf.d files for a supervise-daemon-managed service | ✓ | ✓ | ✓ | ✓ |
| system-fastflowlm | Reuse the system <pkg>sci-ml/fastflowlm</pkg> (/usr/bin/flm) for the FastFlowLM NPU backend instead of letting lemond auto-download the flm runtime at first NPU use | ✓ | ✓ | ✓ | ✗ |
| system-kokoro | Reuse the system <pkg>sci-ml/kokoros</pkg> (/usr/bin/koko) for the Kokoro TTS backend instead of fetching an upstream prebuilt binary at runtime | ✓ | ✓ | ✓ | ✓ |
| system-llamacpp | Reuse the system <pkg>sci-misc/llama-cpp</pkg> (/usr/bin/llama-server) for the llama.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes) | ✓ | ✓ | ✓ | ✓ |
| system-rocm | Point lemond's ROCm runtime at the system ROCm in /usr (via ROCM_PATH) so its rocm-stable image/audio backends reuse it instead of downloading AMD's multi-GB TheRock runtime; requires a system ROCm (pulls <pkg>dev-util/hip</pkg>). Mutually exclusive with system-therock | ✓ | ✓ | ✓ | ✗ |
| system-sdcpp | Reuse the system <pkg>sci-misc/stable-diffusion-cpp</pkg> (/usr/bin/sd-server) for the stable-diffusion.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes) | ✓ | ✓ | ✓ | ✓ |
| system-therock | Point lemond's ROCm runtime at <pkg>dev-util/therock-bin</pkg> in /opt/therock-bin (via ROCM_PATH) instead of downloading AMD's TheRock runtime; pinned to the 7.13 line whose version matches lemond's expected ROCm. Mutually exclusive with system-rocm | ✓ | ✓ | ✓ | ✗ |
| system-whispercpp | Reuse the system <pkg>app-accessibility/whisper-cpp</pkg> (/usr/bin/whisper-server) for the whisper.cpp backend instead of fetching an upstream prebuilt binary at runtime; build it with your preferred GPU backend (see the post-install notes) | ✓ | ✓ | ✓ | ✓ |
| systemd | Keep upstream's systemd units (lemond.service, system + user), running as a dedicated <pkg>acct-user/lemonade</pkg> | ✓ | ✓ | ✓ | ✓ |
| tauri | Build the Tauri desktop app (lemonade-app) -- a native webkit2gtk window wrapping the same renderer as USE=webui. Pulls a Rust toolchain (<pkg>dev-lang/rust</pkg>) plus <pkg>net-libs/webkit-gtk</pkg>, and fetches Rust crates over the network at build time | ✓ | ✓ | ✓ | ✓ |
| webui | Build the bundled React web frontend (served by lemond at the server root) via <pkg>net-libs/nodejs</pkg>; runs npm/webpack over the network at build time. Without it lemond serves a static placeholder page | ✓ | ✓ | ✓ | ✓ |
| Type | File | Size | Versions |
|---|---|---|---|
| DIST | lemonade-11.5.1.tar.gz | 3392860 bytes | 11.5.1 |
| Type | File | Size |
|---|---|---|
| DIST | lemonade-10.10.0.tar.gz | 2778779 bytes |
| DIST | lemonade-11.0.0.tar.gz | 3151240 bytes |