--- title: Train Models Faster with JAX and MaxText Using NVFP4 on... description: NVFP4 is a 4-bit mixed-precision training format from NVIDIA that delivers 1.31–1.73x speedup over FP8 baselines on Blackwell hardware with no measurable... canonical: https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell | daily.dev og:description: NVFP4 is a 4-bit mixed-precision training format from NVIDIA that delivers 1.31–1.73x speedup over FP8 baselines on Blackwell hardware with no measurable... og:url: https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn og:image: https://api.daily.dev/og/posts/OuYTd5oYN.png og:image:alt: Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell 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. # Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell **[NVIDIA Developer](https://daily.dev/sources/nvidiadev)** · 7 min read · 0 upvotes · 0 comments ## Summary NVFP4 is a 4-bit mixed-precision training format from NVIDIA that delivers 1.31–1.73x speedup over FP8 baselines on Blackwell hardware with no measurable accuracy loss. The recipe combines five key techniques: 16-element micro block scaling, E4M3 block scale factors, Random Hadamard Transform on WGRAD inputs, 2D weight scaling, and stochastic rounding. NVFP4 is applied only to MLP layers (not attention) since MLPs account for most training FLOPs. The MaxText framework provides a ready-to-use recipe for Llama 3 8B and 405B pretraining via a single quantization flag. On GB300, Llama 3.1 405B achieves 3633 TFLOPs/GPU vs 2103 for FP8, a 1.73x gain, while training loss curves remain virtually identical. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Using NVFP4 Low-Precision Model Training for Higher Throughput Without Losing Accuracy](https://daily.dev/posts/using-nvfp4-low-precision-model-training-for-higher-throughput-without-losing-accuracy-mburtmoay) · NVIDIA Developer · 1 upvotes · 0 comments - [3 Ways NVFP4 Accelerates AI Training and Inference](https://daily.dev/posts/3-ways-nvfp4-accelerates-ai-training-and-inference-2zfw3y2wb) · NVIDIA Developer · 1 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm) [View this post on daily.dev](https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn) ```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":"Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell","url":"https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn"},"datePublished":"2026-06-08T18:19:41.319Z","dateModified":"2026-06-08T18:39:13.372Z","description":"NVFP4 is a 4-bit mixed-precision training format from NVIDIA that delivers 1.31–1.73x speedup over FP8 baselines on Blackwell hardware with no measurable...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a99694e6da43af30577be52ea52828f1?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a99694e6da43af30577be52ea52828f1?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"NVIDIA Developer","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":"NVIDIA Developer","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/86e45aab42ba48ce83103d01b1119910","url":"https://daily.dev/sources/nvidiadev"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/train-models-faster-with-jax-and-maxtext-using-nvfp4-on-nvidia-blackwell-ouytd5oyn","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"NVIDIA Developer","item":"https://daily.dev/sources/nvidiadev"},{"@type":"ListItem","position":3,"name":"Train Models Faster with JAX and MaxText Using NVFP4 on NVIDIA Blackwell"}]} ```