@zhijianliu_: This is what DFlash was built for. Our block-diffusion drafter + KV injection, now running at frontier scale — thanks t…
Summary
DFlash, a block-diffusion drafter with KV injection, is now running at frontier scale, achieving up to 4.3x greater throughput over baseline, integrated with Modal and SGLang for Qwen 397B.
View Cached Full Text
Cached at: 06/16/26, 03:37 PM
This is what DFlash was built for. ⚡
Our block-diffusion drafter + KV injection, now running at frontier scale — thanks to @modal and @sgl_project for the engine + integration support!
Modal (@modal): We worked with @lmsysorg and https://t.co/Cg0JsVomui to
- integrate DFlash spec into @sgl_project
- make it faster with overlap
- train a DFlash drafter for @Alibaba_Qwen 397B-A17B
The result: up to 4.3x greater throughput over baseline and 1.5x over native MTP.
Similar Articles
@zhijianliu_: DFlash is now running in a production inference stack. More draft models coming soon. https://github.com/z-lab/dflash
DFlash is a lightweight block diffusion model for speculative decoding, now running in production with support for various LLMs like Qwen and Gemma.
DFlash and Spec V2 Decoding (14 minute read)
Z Lab, SGLang, and Modal release DFlash, a new speculative decoding model for Qwen 3.5 397B-A17B that uses block diffusion and KV injection to achieve over 4x throughput improvement over baseline and 1.5x over native MTP.
DFlash: Block Diffusion for Flash Speculative Decoding
DFlash is a new speculative decoding framework that uses a lightweight block diffusion model for parallel token drafting, achieving over 6x acceleration compared to autoregressive methods. It significantly outperforms existing state-of-the-art methods like EAGLE-3 while maintaining high output quality.
z-lab/Qwen3.6-27B-DFlash
This article introduces Qwen3.6-27B-DFlash, a specialized drafter model for DFlash, a novel speculative decoding method using block diffusion to accelerate inference speed. It provides installation instructions for vLLM and SGLang to enable parallel drafting with the target Qwen3.6-27B model.
z-lab/Qwen3.6-35B-A3B-DFlash
z-lab releases DFlash, a speculative decoding drafter that uses a lightweight block-diffusion model to draft 15–16 tokens in parallel, yielding up to 2.9× speedup for Qwen3.6-35B-A3B inference.