/Tech5h ago

SGLang integrates Brian Chao's SPEED algorithm, achieving a 2x speedup for FLUX and other diffusion models with no quality loss

The algorithm progressively increases image resolution during denoising steps.

7594199.7K
Banghua Zhu
Lianmin Zheng#1126
Original post
Brian Chao@BrianCChao

Aside from the official code release, I am thrilled to share that Spectral Progressive Diffusion a.k.a. SPEED (https://arxiv.org/abs/2605.18736) is now integrated into @lmsysorg's SGLang (@sgl_project)! 🚀

Instead of always running diffusion at full resolution, SPEED progressively grows resolution across denoising steps, drastically cutting token count and achieving >2× speedup with no quality loss.

SPEED is now supported in SGLang for FLUX.1 & 2, Z-Image, Qwen-Image, and Wan. Support for Ideogram 4 incoming.

Try it out now: https://docs.sglang.io/docs/sglang-diffusion/progressive_resolution.

[1/4]

Howard Xiao@howard_xhc

Today we release the code and a demo for our recent Spectral Progressive Diffusion paper🎉 Play around with it anytime! Just as what we have been doing also, we hope that it encourages the integration of our plug-and-play framework into latest and greatest image and video generation models!! We also included an agent skill wrapper in the repo, to making things easier.

🛜Project website: https://howardxiao.ca/speed/ 📄Paper: https://arxiv.org/abs/2605.18736 💻Github (with ComfyUI): https://github.com/howardhx/speed 🤗Demo (HuggingFace): https://huggingface.co/spaces/howardhx/speed

9:06 AM · Jun 11, 2026 · 6K Views
Sentiment

Users are excited about SGLang integrating SPEED for over 2x faster diffusion inference because of the open source collaboration and performance gains.

Pos
100.0%
Neg
0.0%
2 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS329
Byron Hsu@hsu_byron

@lmsysorg @sgl_project @BrianCChao cc @jfischoff @isidentical @gorkem may be interested

4hViews 329Likes 1
BOOKMARKS2
Banghua Zhu@BanghuaZ

Great efforts speeding up diffusion model inference to next level!

Brian Chao@BrianCChao

Aside from the official code release, I am thrilled to share that Spectral Progressive Diffusion a.k.a. SPEED (https://arxiv.org/abs/2605.18736) is now integrated into @lmsysorg's SGLang (@sgl_project)! 🚀

Instead of always running diffusion at full resolution, SPEED progressively grows resolution across denoising steps, drastically cutting token count and achieving >2× speedup with no quality loss.

SPEED is now supported in SGLang for FLUX.1 & 2, Z-Image, Qwen-Image, and Wan. Support for Ideogram 4 incoming.

Try it out now: https://docs.sglang.io/docs/sglang-diffusion/progressive_resolution.

[1/4]

1hViews 318Likes 2Bookmarks 2
LIKES3
Kyle Kranen@KranenKyle

@hsu_byron @lmsysorg @sgl_project @BrianCChao Being humble that YOU were the one to add PD to SGL 😉

4hViews 95Likes 3
RETWEETS1
Byron Hsu@hsu_byron

An absolutely impressive contribution to SGLang (@lmsysorg @sgl_project) by @BrianCChao, providing a 2× speedup with no loss in quality!

In my opinion, there are three levels of contributions one can make to SGLang. The first is fixing bugs or extending existing features, such as fused kernels. The second is adding a new feature based on an existing paper, such as PD disaggregation. The third is inventing a new algorithm and integrating it into SGLang.

This work clearly falls into the third category, and I’m excited to hear feedback from the community!

Thanks a lot to the review from @mick_qian and supports from @ying11231 @BanghuaZ @lm_zheng

Brian Chao@BrianCChao

Aside from the official code release, I am thrilled to share that Spectral Progressive Diffusion a.k.a. SPEED (https://arxiv.org/abs/2605.18736) is now integrated into @lmsysorg's SGLang (@sgl_project)! 🚀

Instead of always running diffusion at full resolution, SPEED progressively grows resolution across denoising steps, drastically cutting token count and achieving >2× speedup with no quality loss.

SPEED is now supported in SGLang for FLUX.1 & 2, Z-Image, Qwen-Image, and Wan. Support for Ideogram 4 incoming.

Try it out now: https://docs.sglang.io/docs/sglang-diffusion/progressive_resolution.

[1/4]

4hViews 3.8KLikes 30Bookmarks 10
REPLIES1
WK@zoom_will

@BrianCChao @lmsysorg @sgl_project Speed is opensource? Can we make Ideogram 4.0 + SPEED?

5hViews 49Likes 1
Brian Chao@BrianCChao

SPEED achieves training-free diffusion acceleration, preserving high image and video quality across all tested models with significant efficiency boost.

[2/4]

6hViews 206Likes 2Bookmarks 1
Brian Chao@BrianCChao

Extensive benchmarking results show SPEED achieves up to 2× speedup across various image and video models. Our method is robust: it requires only a single tuning parameter δ that trades off image quality against speedup, where larger δ corresponds to more coarse-resolution stages and larger speedup.

[3/4]

6hViews 146Likes 1
Brian Chao@BrianCChao

Many thanks to @hsu_byron for the connection and immense support from @mick_qian and the @sgl_project team. Super excited to be doing open source again, and to bring our research directly into production!

SPEED is a joint work led by @howard_xhc, with @YarivLior and @GordonWetzstein.

🌐 Website: https://howardxiao.ca/speed/ 💻 Code: https://github.com/howardhx/speed 📄 Paper: https://arxiv.org/abs/2605.18736 🚀 Demo: https://huggingface.co/spaces/howardhx/speed

Check out our original paper post:

6hViews 216Likes 3
Brian Chao@BrianCChao

@hsu_byron @lmsysorg @sgl_project 🚀🚀🚀

4hViews 93
Brian Chao@BrianCChao

@zoom_will @lmsysorg @sgl_project Ideogram 4 SGLang support is coming soon! We will release a ComfyUI node for Ideogram as well.

4hViews 43