qwen-image-2.1-viggle-turbo: Viggle/Qwen‑Image‑2.1‑viggle‑turbo: 6‑step distilled text‑to‑image model

• Fact Model is a 6‑step distilled version of Qwen‑Image‑2.1[1]AIA few-step distilled version of Qwen/Qwen-Image-2.1 by Viggle.
• Fact Runs 5× faster than the 40‑step base model end‑to‑end[1]AIAbout 5× faster than the 40-step base model end to end
• Fact No classifier‑free guidance is used[1]AIwith no classifier-free guidance
• Fact 9‑step mode adds two extra steps after LoRA is switched off[1]AINew 9-step mode: 7 turbo steps, then the LoRA is switched off and the base model finishes the last two
• Fact 6‑step configuration uses specific sigmas and CFG scale[1]AI6 steps with sigmas=[1.0, 0.9375, 0.875, 0.75, 0.5, 0.25], truecfgscale=1.0, no negative prompt
• Fact Viggle/Qwen-Image-2.1-viggle-turbo model card is listed among trending models on the Hugging Face Hub.[1]
• Fact That is 530 likes in 11 days, an average of 49.9 per day.[1]

• Analysis Use the 6‑step pipeline for speed, or 9‑step for higher detail; adjust sigmas and CFG as described; load LoRA via peft; consider ComfyUI single‑file transformers for deployment.AI
• Analysis Automatically detected from public activity data. Not yet reviewed by an analyst.

• Fact Model is a 6‑step distilled version of Qwen‑Image‑2.1[1]AIA few-step distilled version of Qwen/Qwen-Image-2.1 by Viggle.
• Fact Runs 5× faster than the 40‑step base model end‑to‑end[1]AIAbout 5× faster than the 40-step base model end to end
• Fact No classifier‑free guidance is used[1]AIwith no classifier-free guidance
• Fact 9‑step mode adds two extra steps after LoRA is switched off[1]AINew 9-step mode: 7 turbo steps, then the LoRA is switched off and the base model finishes the last two
• Fact 6‑step configuration uses specific sigmas and CFG scale[1]AI6 steps with sigmas=[1.0, 0.9375, 0.875, 0.75, 0.5, 0.25], truecfgscale=1.0, no negative prompt

Architecture

Insufficient evidenceNo architecture has been documented for this signal.

Insufficient evidenceDeveloper implications have not been assessed yet.

Insufficient evidenceNo adoption indicators recorded.

Timeline

1. 2 Oct 2026

   First seenFirst seen by the Astrava pipeline[1]

• Fact Small, dense text and complicated edits are weaker than the base model[1]AIThe clearest gaps are small, dense text and complicated edits (Known limitations)
• Fact Finer detail and small text accuracy improve only in 9‑step mode[1]AIFiner detail, and small text comes out right more often (not always)
• Analysis Generated from activity metrics only; verify the linked sources before acting on it.

Insufficient evidenceNo competing approaches documented yet.

Insufficient evidenceNo independent projects or organisations are documented as building on this yet.