Lemonade

Lemonade is a local AI server for people who want to chat with models on their own hardware or use them in apps and agents. It includes a built-in interface for direct use and APIs for connecting other software.

• Chat and vision: Use models for conversations, coding and tasks that involve images, with embeddings available for applications that need them.
• Images and speech: Generate images and spoken audio, or transcribe speech using local models such as Stable Diffusion, Kokoro and Whisper.
• App connections: Connect clients through OpenAI, Anthropic and Ollama compatible APIs, or bundle the embeddable edition into your own application so users don't need a separate Lemonade installation.
• Model choice: Browse and download models with the Model Manager, including GGUF, FLM and ONNX formats, and bring in custom GGUF or ONNX models from Hugging Face or ModelScope.

Lemonade is free and open source under Apache 2.0. It runs on Windows, macOS and Linux, with Docker also available. Depending on the model and backend, inference can use a CPU, supported NVIDIA or AMD GPUs, Apple Silicon GPUs, or AMD XDNA2 NPUs. Hardware support varies by operating system and capability.

Local inference keeps processing on your hardware, and Lemonade sends no telemetry. Optional routing can send requests to servers or cloud APIs instead. Requested model downloads and registry lookups can contact Hugging Face or ModelScope.