Amaretto is a line of open-weight models, trained with care, released with full weights, and sized for the hardware people actually have.
The line
A model for literary writing and writing assistance, tone adjustment, editing, summarization and style transfer. Runs on everyday hardware via GGUF quantizations, from llama.cpp to Ollama. Best for: chat, editing, style transfer.
The house standard. Literary prose with more depth: scene writing, character voice, and emotional interiority, alongside editing, tone adjustment, and style transfer. GGUF quantizations to fit a single consumer GPU. Best for: chat, scene composition, character voice, style transfer, editing, summarization.
The top of the line. The most capable model in the line, built for creative-heavy work and long documents where the smaller ones fall short. Best for: chat, scene composition, character voice, style transfer, editing, analysis, summarization.