<h1 id="pruna-qwen-image-2.1">Pruna-Qwen-Image-2.1</h1>Unofficial mirror. This listing mirrors the original project for reference and convenience. It is not created, maintained, or endorsed by PrunaAI. Please credit the original authors and preserve all license, attribution, and modification notices.Official repositories: <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1">PrunaAI/P-Q-Image-2.1</a> · <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1-Comfy">Official ComfyUI pack and workflows</a> Base model: <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-2.1">Qwen/Qwen-Image-2.1</a><h2 id="overview">Overview</h2>These LoRA adapters speed up text-to-image generation and image editing with Qwen-Image-2.1. They are not standalone models; the Qwen-Image-2.1 base checkpoint is required.<h2 id="v0.2-6-step-adapters">v0.2 — 6-step adapters</h2>Two alternatives are available. Load only one at a time.6-steps: <code>p_q_image_2.1_6_steps_v0.2_comfyui_bf16.safetensors</code> Faithful base-model behavior; especially useful for editing where the input should be preserved.6-steps-upsampled: <code>p_q_image_2.1_6_steps_upsampled_v0.2_comfyui_bf16.safetensors</code> Improved results from shorter prompts without a separate prompt-upsampling call.The upsampled adapter may not reproduce exactly what the original prompt upsampler would produce.<h2 id="v0.1-existing-versions">v0.1 — existing versions</h2>The 8-step and 5-step adapters remain available. The 8-step version favors quality; the 5-step version is faster but has lower visual quality. Each adapter has its own sampling schedule.<h2 id="comfyui-settings">ComfyUI settings</h2>Use an updated ComfyUI version with native Qwen-Image-2.1 support. Load the adapter with the built-in <code>LoraLoaderModelOnly</code> node on the Qwen-Image-2.1 base checkpoint, not the Turbo checkpoint.For either v0.2 adapter, use Euler, CFG 1, and LoRA strength 1.0. Set <code>ManualSigmas</code> to:<code>1.0, 0.961054, 0.918949, 0.769302, 0.545554, 0.222292, 0</code>Do not use the v0.1 schedule with a v0.2 adapter. No negative prompt or custom nodes are required.<h2 id="prompts-and-resolution">Prompts and resolution</h2>Start at 1024 × 1024. The adapters were trained at 1K resolution for text-to-image and editing, with up to three reference images. Higher resolutions, including 2K, and additional reference images are outside the training coverage, so results may vary. Detailed prompts are still useful; for the upsampled adapter, start with your original prompt.<h2 id="license-and-attribution">License and attribution</h2>This is a derivative of Qwen-Image-2.1 and is subject to the <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-2.1/blob/main/LICENSE">Qwen Research License Agreement</a>. Review the license before use or redistribution. This mirror grants no additional rights; preserve the original attribution and license notices.

Turbo - Few-step LoRA adapters - Pruna Qwen Image 2.1

p_qwen_image_2.1_8step_v0.1.safetensors

pruna 8 steps v0.1

pruna 5 steps v0.1

<h1 id="p-q-image-2.1">P-Q-Image-2.1</h1>Official listing from Pruna AI. Same LoRA adapters as <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1">PrunaAI/P-Q-Image-2.1</a>.<hr />Few-step LoRA adapters for Qwen-Image-2.1P-Q-Image-2.1 is a set of LoRA adapters that let <a target="_blank" rel="ugc" href="https://huggingface.co/Qwen/Qwen-Image-2.1">Qwen/Qwen-Image-2.1</a> generate and edit images in only 6 steps (v0.2), or 5 or 8 steps (v0.1). The adapters load on top of the base model, so the pipeline, text encoder, and VAE stay unchanged.v0.2: two new 6-step adapters.<ul><li>An advanced distillation for more faithful generation and editing.</li><li>The first distillation removing need for prompt upsampling into the LoRA, thus significantly reducing end-to-end generation.</li></ul>The v0.1 adapters remain available.<h2 id="6-step-variants">6-step variants</h2>Pick 6-steps for faithful base-model behavior, or 6-steps-upsampled for improved visual quality from a short prompt, without a separate prompt-upsampling call.| Variant | File | Steps | Trade-off ||---|---|:---:|---|| <code>6-steps</code> | <code>p_q_image_2.1_6_steps_v0.2.safetensors</code> | 6 | Advanced distillation. Recommended for preserving inputs when editing. || <code>6-steps-upsampled</code> | <code>p_q_image_2.1_6_steps_upsampled_v0.2.safetensors</code> | 6 | Improved visual quality, shorter end-to-end generation. The first distillation that removes the need for a separate prompt-upsampling step. |6-steps is an advanced 6-step distillation of the base model. It is intended as a drop-in replacement for the base model in existing generation and editing workflows, using the adapter's step count and sigma schedule. It improves on the v0.1 adapters in almost all categories and is particularly good at editing tasks where preserving the input matters.6-steps-upsampled distills the prompt upsampler recommended for Qwen-Image-2.1 into a 6-step LoRA. It improves visual quality without a separate prompt-upsampling call, which significantly reduces end-to-end generation time. It may not reproduce exactly what the actual prompt upsampler would do, and we will continue improving it.<h3 id="v0.1">v0.1</h3>Both adapters are v0.1. Pick one based on whether you need quality or speed.| Variant | File | Steps | Trade-off ||---|---|:---:|---|| <code>8-steps</code> | <code>p_q_image_2.1_8_steps_v0.1.safetensors</code> | 8 | Higher quality. Recommended as the default. || <code>5-steps</code> | <code>p_q_image_2.1_5_steps_v0.1.safetensors</code> | 5 | Higher speed, with noticeably lower visual quality. |Each adapter is trained for its own sigma schedule. Load only one at a time. For the v0.2 adapters, do not reuse the v0.1 schedules.<h2 id="prompts-editing-and-resolution">Prompts, editing, and resolution</h2>For 6-steps, use the same prompts and reference images you would use with the base model. For 6-steps-upsampled, start with your original prompt; the adapter is intended to capture the benefit of prompt upsampling during generation. Detailed prompts still let you specify the subject, setting, lighting, and style.The adapters were trained at 1K resolution only, with a mix of simple and upsampled prompts, text-to-image generation, and single- and multi-image editing with up to 3 reference images. Prompt upsampling is optional; detailed prompts usually give better results.Start at 1024 × 1024 and use at most 3 reference images for editing. Higher resolutions (including 2K) and more reference images may work, but are outside the training coverage, and quality may vary. The 2K timings below show inference speed, not a guarantee of quality at that resolution.<h2 id="quickstart-and-benchmark">Quickstart &amp; Benchmark</h2>Check on: <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1">PrunaAI/P-Q-Image-2.1</a>.<h2 id="limitations">Limitations</h2>- The <code>6-steps-upsampled</code> adapter can improve visual quality without a separate prompt-upsampling call, but may not exactly reproduce the output of the actual prompt upsampler.- <code>6-steps</code> is intended to preserve the base model's behavior in fewer steps; distillation does not guarantee identical outputs.- Use the step count and sigma schedule that match your v0.2 adapter. Other step counts, schedules, or CFG values are not supported.&lt;details&gt;&lt;summary&gt;&lt;b&gt;Known limits of the 5–8 step adapters (v0.1)&lt;/b&gt;&lt;/summary&gt;<h3 id="v0.1">v0.1</h3>- This is a first version. Quality is below that of the base model.- The 5-step adapter is faster than the 8-step one, but its images are visibly worse.- Short or vague text-to-image prompts give weaker results.- The adapters are only intended for the settings above. Other step counts, schedules, or CFG values are not supported.<h3 id="what-this-is-not">What this is not</h3>- Not a standalone model. It needs the <code>Qwen/Qwen-Image-2.1</code> base weights.- Not a replacement for the base model when you need its full quality.- Not a finished release. Expect the weights to change in future versions.<h2 id="license">License</h2>The end distilled model is based on Qwen-Image-2.1.P-Q-Image-2.1 is a derivative of Qwen-Image-2.1 and is distributed under the Qwen RESEARCH LICENSE AGREEMENT. Review its use restrictions before using or redistributing the adapters. See the live <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1/blob/main/LICENSE">LICENSE</a> and <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI/P-Q-Image-2.1/blob/main/NOTICE">NOTICE</a>.Qwen is licensed under the Qwen RESEARCH LICENSE AGREEMENT, Copyright (c) 2026 Hangzhou Tongyi Laboratory Technology Co., Ltd. All Rights Reserved.<h2 id="what's-next">What's next?</h2>- Use P-Q-Image-2.1 to generate and edit images with Qwen-Image-2.1 in a few steps.- Compress your own models with <a target="_blank" rel="ugc" href="https://github.com/PrunaAI/pruna">Pruna</a> and give us a ⭐️ for more efficiency!

P-Q-Image-2.1

Today, we open-source v0.2 of P-Q-Image-2.1, two new 6-step LoRA adapters for Qwen-Image-2.1 by Qwen. The first distillation that removes the prompt-upsampling step!- 𝗛𝗼𝘄 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝘁 𝗶𝘀 𝗶𝘁? Generate or edit images in just 6 steps instead of 40. Use 6-steps for preserving inputs when editing and 6-steps-upsampled for improved visual quality and shorter end-to-end generation.- 𝗛𝗼𝘄 𝗴𝗼𝗼𝗱 𝗶𝘀 𝗶𝘁? 6-step distillation provides improved quality with few generation steps. 6-steps-upsampled is the first distillation that removes the need for a separate prompt-upsampling step. This means 41 to 80 seconds saved, depending on your prompt style.- 𝗛𝗼𝘄 𝗱𝗼𝗲𝘀 𝗶𝘁 𝘄𝗼𝗿𝗸? The LoRA adapters load directly on top of Qwen-Image-2.1, so the pipeline stays unchanged. Few-step distillation cuts the diffusion cost, and the upsampled adapter also removes the prompt-upsampling step before image generation.

p_q_image_2.1_6_steps_v0.2.safetensors

p_q_image_2.1_6_steps_upsampled_v0.2.safetensors

p_q_image_2.1_5_steps_v0.1.safetensors

p_q_image_2.1_8_steps_v0.1.safetensors

v0.2

PrunaAI/isHeSatoshi turbo LoRAs can be used for editing/text-to-image Current Viggle turbo LoRAs can be used for text-to-image Current Best settings for Viggle/PrunaAI 4-step: er_sde/beta res_2s_ode/bong_tangent, 5-8 steps, cfg 1-2 (+ lenovo/zjourney refiner lora to remove some of the overbaked, high-contrast, yellow tint) isHeSatoshi 6-step: res_2s_ode/ManualSigmas(1.0, 0.9375, 0.875, 0.75, 0.5, 0.25, 0.0) All examples are generated with 1.0 strength I did not train, distill, or create the original Turbo LoRA weights. This model page provides only modified compatibility versions intended to let the remaining compatible LoRA adapters load with ComfyUI's built-in LoRA loader Full credit for the original Qwen-Image-2.1-viggle-turbo-4step-lora-r64.safetensors belongs to the <a target="_blank" rel="ugc" href="https://huggingface.co/Viggle">ViggleAI</a> Team. Full credit for the original p_qwen_image_2.1_5/8step_v0.1.safetensors belongs to the <a target="_blank" rel="ugc" href="https://huggingface.co/PrunaAI">PrunaAI</a> Team. Full credit for the original Qwen-Image-2.1-turbo-v0.2.1-6step-lora-r128s belongs to <a target="_blank" rel="ugc" href="https://huggingface.co/isHeSatoshi">isHeSatoshi</a>. Viggle statement for v0.1: "Status: v0.1 preview, work in progress — this release still falls short of the base model. On complicated image editing (multi-reference composition, face swaps, identity-preserving edits, instructions with several constraints) it is clearly worse than the 40-step base model. Text-to-image at 4 steps is usable. We are still working on it and will update this repository as the distillation improves; treat the current weights as a preview, not a replacement for the base model." PrunAI statement for v0.1: "Pruna-Qwen-Image-2.1 is a set of LoRA adapters that let Qwen-Image-2.1 generate and edit images in only 5 or 8 steps. The adapters load on top of the base model, so the pipeline, text encoder, and VAE stay unchanged. Training is based on DMD. Improved using Qwen.v0.1: first release, work in progress. Pruna-Qwen-Image-2.1 does not yet match the visual quality of the base model. We are still improving the distillation and will update this repository as new versions become available."

Qwen 2.1 Turbo LoRas

pruna_qwen_2.1_8step_v0.1

qwen_2.1_turbo_extract

isHeSatoshi_6-stepLoRA

pruna_qwen_2.1_5step_v0.1

Qwen_Viggle_4step_0.1

### 当前模型的贡献者未提供更加详细的模型介绍。模型文件和权重，可浏览“模型文件”页面获取。
#### 您可以通过如下git clone命令，或者ModelScope SDK来下载模型

SDK下载
```bash
#安装ModelScope
pip install modelscope
```
```python
#SDK模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('bozoyan/p_q_image_2.1_8_steps')
```
Git下载
```
#Git模型下载
git clone https://www.modelscope.cn/bozoyan/p_q_image_2.1_8_steps.git
```

如果您是本模型的贡献者，我们邀请您根据<a href="https://modelscope.cn/docs/ModelScope%E6%A8%A1%E5%9E%8B%E6%8E%A5%E5%85%A5%E6%B5%81%E7%A8%8B%E6%A6%82%E8%A7%88" style="color: lightgrey; text-decoration: underline;">模型贡献文档</a>，及时完善模型卡片内容。

Files with SHA256: f0865d68b02511a3a0ed232d9d1aa99cac3a94166f574b38bcbab1a7a297bb15