Subtopic Small Models · Vision Language · Quantization Takeaways − Mitsuba is a ternary 1.58-bit quantization of Qwen3.8-27B, shrunk to 7.3 GB for a single 16 GB GPU. Purpose-built for ComfyUI: image to prompt generation for Stable Diffusion, Krea, and video pipelines.
Alibaba s Qwen team shipped Qwen-Image-2.1 on September 20, 2026, and it changes the calculus for anyone running AI image generation on their own hardware in Australia. The 20-billion-parameter model handles text-to-image generation, multi-reference editing with up to 10 source images, and native transparent PNG output from a single checkpoint.
Housekeeping: Paid subscribers to Interconnects now get a permanent 40% discount on my book when purchasing at Manning.com.
Botika builds agentic e-commerce teams that automate visual production for global fashion brands. Powered by custom models researched and trained in-house, Botika handles the entire pipeline - from 4K image generation to real-time personalization at scale.
Comfy Org is turning one of AI image generation's most flexible open-source tools into something far easier to install, manage and use at work. ComfyUI has always had a powerful pitch: build visual AI workflows exactly the way you want them, node by node, model by model, without waiting for a closed platform to decide what matters.
ComfyUI hits $500M valuation as creators seek more control over AI-generated media | TechCrunch Last day to exhibit your breakthrough to 10,000+ tech leaders at Disrupt is on Oct 2 . Book Exhibit Table Now.