Google 发布 DiffusionGemma：文本扩散模型，推理速度提升 4 倍
推理速度提升 4 倍意味着更低的延迟和更低的成本，做实时文本生成或大规模内容生产的团队值得关注这个新方向。
Google 发布了 DiffusionGemma，这是一个基于 Gemma 4 的实验性开放模型，采用文本扩散技术。
与传统逐词预测不同，DiffusionGemma 能同时生成整段文本，实现高达 4 倍的推理加速。该模型目前以研究预览形式开放，旨在探索更高效的文本生成方式。
这对于需要低延迟文本生成的场景（如实时对话、内容创作）具有重要意义。
DiffusionGemma is an open, experimental model that brings our text diffusion research to Gemma 4. It’s a racehorse 🏇achieving up to 4x faster inference by generating entire blocks of text simultaneously vs predicting token-by-token (word-by-word) output!
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include 文本扩散, 推理加速, Gemma 4, 开放模型.
Source attribution: curated from Sundar Pichai (@sundarpichai). Cross-source corroboration: 8 independent sources have covered this development (including LMSYS Org (SGLang), Philipp Schmid, vLLM), confirming industry-wide signal strength.
