unsloth ships FP8-quantized Qwen-Image-2.1 for cheaper image generation and editing
unsloth · hf · 2026-09-21
unsloth released Qwen-Image-2.1-FP8 on Hugging Face, an FP8/INT8 quantized build of the Qwen-Image text-to-image model, compatible with diffusers.
It supports text-to-image generation, image editing, and RGBA output, enabling deployment on lower-VRAM setups.
Related event: Unsloth's quantized Qwen-Image-2.1 runs locally on just 6GB VRAM(2 posts)→
More from Multimodal
- Two-Ref Workflow: Character Sheet + Real Shot with GPT Image 2.5 and Seedance 2.5 — techhalla · 2026-09-23
- Qwen-Image-2.1 official demo live: 40 steps plus PE thinking unlocks full power — _akhaliq · 2026-09-23
- AntLing open-sources Ming-Image-0.1-Design, tops open-weight UI design leaderboard — niacolhealth · 2026-09-23
- Qwen Image 2.1 tested: solid at 40 steps with CFG 1, but a yellow tint remains — rm_rf_all_files · 2026-09-23
- Mirage launches Tesseract, a video creative suite built for AI agents — _AustinCalvert_ · 2026-09-23
- MiniMax H3 generates cinema-grade title sequence from a detailed one-shot prompt — umesh_ai · 2026-09-23