Qwen-Image-2.1-Turbo cuts 7B image model to 8 denoising steps
lmoroney · x · 2026-10-10
- Qwen released Qwen-Image-2.1-Turbo on Hugging Face, an accelerated checkpoint of the 7B Qwen-Image-2.1 supporting text-to-image generation and image editing.
- It runs in 8 denoising steps (vs 40 in the base model's examples), ships with a built-in sampling schedule, defaults to CFG=1, and uses prefix KV caching to reuse text and reference-image context across steps.
- Requires the latest Diffusers source (with PR #14950) and transformers>=5.17.0; loads via QwenImage21Pipeline in bfloat16. The author notes one detail in the 8-step setup that will trip people up.
More from Multimodal
- Testing spatial-temporal consistency: recreating one moment from two views on Kling 4.0 Flash — umesh_ai · 2026-10-11
- Open-source art animation skill offers 35 art styles and 9 narration grammars — AlchainHust · 2026-10-11
- Qwen-Image Edit lands native support in ComfyUI — saroxel · 2026-10-11
- AI video made with Seedance 2.5 hailed as a masterpiece — SimplyAnnisa · 2026-10-11
- Anthropic launches Claude Motion beta: turn reports into editable MP4 animated explainers — PrajwalTomar_ · 2026-10-11
- Google and Tel Aviv researchers unveil SepGen, generating video with per-source stems for 4D spatial audio — YonatanBitton · 2026-10-11