Run open-weight image models like Krea 2 on your own GPU — a polished app that just works. No Python, no nodes, no guesswork.
TinyPine keeps the stable part — the inference engine — and wraps it in a thin app that owns its own runtime, picks the right model for your hardware, and gives feedback on everything.
Download, run, generate. No environments, no dependency hell, no node graphs.
It detects your VRAM and picks the right precision automatically — 16 GB or 24.
Curated one-click model packs — “Pines” — that carry everything a model needs.
Your work stays yours, and the things you install are exactly what they claim to be.
Generation runs on your machine. Nothing leaves it without your say-so.
Every Pine is cryptographically signed and verified before it installs.
Blocks illegal content; keeps your creative freedom otherwise.