--- title: I tried Google's new DiffusionGemma, and watching it... description: Google's DiffusionGemma is an experimental open model that generates text using a diffusion process rather than left-to-right autoregressive token generation.... canonical: https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM | daily.dev og:description: Google's DiffusionGemma is an experimental open model that generates text using a diffusion process rather than left-to-right autoregressive token generation.... og:url: https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm og:image: https://api.daily.dev/og/posts/0OP3vD8NM.png og:image:alt: I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM **[XDA Developers](https://daily.dev/sources/xda-developers)** · 9 min read · 0 upvotes · 0 comments ## Summary Google's DiffusionGemma is an experimental open model that generates text using a diffusion process rather than left-to-right autoregressive token generation. Instead of streaming tokens one at a time, it works on a 256-token canvas, repeatedly refining placeholder text until the answer coalesces — visually similar to watching an image generator denoise. Running it locally requires a custom llama.cpp fork and a dedicated llama-diffusion-cli runner via Unsloth's GGUF path. On an M4 Pro MacBook Pro, it was slower than regular Gemma 4 and caused system-wide slowdowns, as Google's speed claims (up to 4x faster, 700+ tokens/sec) apply to dedicated Nvidia GPUs, not Apple Silicon. Quality-wise, Google acknowledges it trails standard Gemma 4 across benchmarks. The model's appeal is conceptual: bidirectional attention within a block canvas enables inline editing, code infilling, and non-linear text structures that left-to-right generation handles less naturally. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Google’s DiffusionGemma is 4x faster than its other Gemma models](https://daily.dev/posts/google-s-diffusiongemma-is-4x-faster-than-its-other-gemma-models-vjw4zukla) · The New Stack · 2 upvotes · 0 comments - [DiffusionGemma: 4x faster text generation](https://daily.dev/posts/diffusiongemma-4x-faster-text-generation-yowvfywbx) · DeepMind · 1 upvotes · 0 comments - [Google unveils DiffusionGemma, an AI model that breaks free of left-to-right processing](https://daily.dev/posts/google-unveils-diffusiongemma-an-ai-model-that-breaks-free-of-left-to-right-processing-03hzpmlaz) · InfoWorld · 0 upvotes · 0 comments - [NVIDIA Accelerates Google DeepMind’s DiffusionGemma for Local AI](https://daily.dev/posts/nvidia-accelerates-google-deepmind-s-diffusiongemma-for-local-ai-xcxwdxism) · NVIDIA · 0 upvotes · 0 comments --- Tags: [#local-ai](https://daily.dev/tags/local-ai), [#text-generation](https://daily.dev/tags/text-generation), [#gemma](https://daily.dev/tags/gemma), [#llama-cpp](https://daily.dev/tags/llama-cpp) [View this post on daily.dev](https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM","url":"https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm"},"datePublished":"2026-06-17T10:03:24.266Z","dateModified":"2026-06-17T10:04:00.251Z","description":"Google's DiffusionGemma is an experimental open model that generates text using a diffusion process rather than left-to-right autoregressive token generation....","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/3680e7c2d9498935b0f2e85f49a0ff65?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/3680e7c2d9498935b0f2e85f49a0ff65?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"XDA Developers","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"XDA Developers","logo":"https://media.daily.dev/image/upload/s--kCg6yyAP--/f_auto,q_auto/v1774964407/logos/xda-developers?_a=BAMAMiWQ0","url":"https://daily.dev/sources/xda-developers"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/i-tried-google-s-new-diffusiongemma-and-watching-it-generate-text-like-an-image-is-unlike-any-local-0op3vd8nm","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"local-ai,text-generation,gemma,llama-cpp","timeRequired":"PT9M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"XDA Developers","item":"https://daily.dev/sources/xda-developers"},{"@type":"ListItem","position":3,"name":"I tried Google's new DiffusionGemma, and watching it generate text like an image is unlike any local LLM"}]} ```