GitHub radar
Krea 2 Turbo: 12B Image Generator You Can Run Locally
Krea.ai released the weights of Krea 2 Turbo — a 12-billion-parameter text-to-image model that runs on a laptop with 16 GB RAM and is already supported by Draw Things and DiffusionBee.
Krea 2 Turbo is a 12-billion-parameter text-to-image diffusion model from Krea.ai, released with open weights under the Krea 2 Community License. It generates images from natural-language text descriptions and is designed for creative, commercial, and research use cases — from quick concepts to full visual production. The model is already integrated into Draw Things and DiffusionBee (macOS apps for local image generation), as well as Google Colab and Kaggle notebooks. Over 100 community-built Spaces on Hugging Face showcase its capabilities. In 4-bit quantization, it fits in approximately 7.7 GB of memory, making it practical for a laptop with 16 GB RAM.
Why a vibe-coder should care
With open weights, Krea 2 Turbo lets you generate images locally — no subscription, no usage limits, no data sent to a cloud service. If you already use Draw Things or DiffusionBee on a Mac, you can add it there directly. For anyone building visual content workflows with an AI agent, this is a capable local alternative to paid API-based generators.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up Krea 2 Turbo locally for me: https://huggingface.co/krea/Krea-2-Turbo — pick the right quantized version for my Mac, install it via Draw Things or DiffusionBee, and show me how to generate a first image.
Ноутбук от 16 ГБ оперативки — в 4-битной версии модель занимает около 7.7 ГБ.
Open on Hugging Face▌ More finds