Working Group: Cross-Vendor Support

June 4, 2026 ยท View on GitHub

Overview

Lead: This working group is led by Ken VanDine, whose handle is @kenvandine on both GitHub and Discord.

Background: Lemonade supports a range of hardware and operatings systems (collectively referred to as platforms). However, achieving full support across all mass-market platforms requires additional compilation targets, platform-specific backend support, etc.

Why: Developers are far more likely to build on Lemonade if it enables their applications to deploy across all relevant platforms.

Goal: Lemonade works on all mass-market platforms and delivers optimized performance on each.

Contributing

Please see the general contribution guidelines, then contact @kenvandine on Discord to get started.

Maintanance

Please see the contribution guide to see who is responsible for maintaining each backend. Working groups are focused areas of roadmap development, and are a subset of the end-to-end maintainance and support of this project.

Roadmap

Roadmap items may be high-level objectives that may span multiple issues and PRs.

Vendor-Neutral Vulkan

  • LLMs via llama.cpp
WindowsLinux
x86xx
ARM
  • Image generation and editing via stable-diffusion.cpp
WindowsLinux
x86xx
ARM
  • Realtime transcription via whisper.cpp
WindowsLinux
x86x
ARM

AMD ROCm

  • LLMs via llama.cpp
WindowsLinux
x86xx
  • Image generation and editing via stable-diffusion.cpp
WindowsLinux
x86xx
  • Realtime transcription via whisper.cpp
WindowsLinux
x86x

Nvidia CUDA

  • LLMs via llama.cpp
WindowsLinux
x86xx
ARM
  • Image generation and editing via stable-diffusion.cpp
WindowsLinux
x86
ARM
  • Realtime transcription via whisper.cpp
WindowsLinux
x86
ARM

Intel OpenVino

  • LLMs via llama.cpp
WindowsLinux
x86
  • Image generation and editing via stable-diffusion.cpp
WindowsLinux
x86
  • Realtime transcription via whisper.cpp
WindowsLinux
x86

Qualcomm QNN

  • LLMs via llama.cpp
WindowsLinux
ARM
  • Image generation and editing via stable-diffusion.cpp
WindowsLinux
ARM
  • Realtime transcription via whisper.cpp
WindowsLinux
ARM

Apple Silicon Metal

  • LLMs via llama.cpp
  • Image generation and editing via stable-diffusion.cpp
  • Realtime transcription via whisper.cpp