Models & Labs
HP ZGX Fury: GB300 Local Inference at 3k Tokens
Sam Witteveenhigh confidence
Why it matters
- →GB300 chipset enables high-performance local inference without cloud dependency.
- →748GB memory allows running significantly larger models than standard consumer GPUs.
- →Pre-integrated enterprise software stack reduces deployment friction for on-prem AI.
HP has unveiled the ZGX Fury, an AI workstation featuring NVIDIA's GB300 chipset, 748GB of memory, and 20 petaflops of performance. The system includes HP Z Runtime (ZRT) and is Red Hat Enterprise Linux certified, targeting on-premises deployment for enterprise teams. Benchmarks cited in early reviews suggest local inference speeds reaching 3,000 tokens per second. This hardware aims to provide data-center scale capabilities within a single desktop form factor.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1NVIDIA Highlights Performance per Watt for AI Efficiency
NVIDIA Blog · July 14, 2026 · Related
- 2General Compute secures $400M loan for inference chips
TechCrunch AI · July 17, 2026 · Related
- 3Idle GPUs: The New Bottleneck in AI Infrastructure
Hugging Face Blog · July 30, 2026 · Related
- 4NVIDIA Extends Vera Rubin for Agentic AI Inference
NVIDIA Blog · August 24, 2026 · Background
- 5Together AI Launches Preemptible Compute for GPU Clusters
Together AI Blog · September 10, 2026 · Background
- 6Together AI deploys NVIDIA B300 cluster on IBM Cloud
Together AI Blog · October 6, 2026 · Related
- 7MIT documents AI hardware evolution over eight years
MIT News AI · October 6, 2026 · Background
- 8General Compute raises $100M Series A for inference chips
TechCrunch AI · October 27, 2026 · Related
What happened next
- 9Altman argues cloud RAM over local inference
The Verge AI · October 30, 2026 · Background
- 10llama.cpp b12027 adds CUDA 13 and Snapdragon support
llama.cpp Releases · November 1, 2026 · Background
More in Models & Labs
Models & Labsmodels
llama.cpp b12029 adds CUDA 13 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally supporting NVIDIA's latest CUDA 13 stack alongside AMD's ROCm 10.0. The inclusion of native builds for Linux arm64 with Snapdragon chips, covering CPU, Adreno GPU, and Hexagon NPU, signals a serious push into mobile edge deployment. While Apple Silicon KleidiAI is disabled in this specific build, the broader hardware coverage means developers no longer need to compile from source for most modern accelerators. It’s less about new features and more about removing friction for anyone running local models on bleeding-edge or diverse hardware.
llama.cpp Releases
© The Verge AIModels & Labsother
Apple Intelligence arrives in China via iOS 27.3
Apple is finally unlocking its AI suite for Chinese users with the iOS 27.3 beta, marking a significant regulatory milestone after months of approval delays. The rollout relies on Alibaba’s Qwen model rather than OpenAI, integrating features like notification summaries and visual intelligence directly into the OS. By sidestepping the new Siri in favor of established text and image tools, Apple is prioritizing immediate utility over the full generative experience. This move effectively brings Apple’s AI capabilities to a massive new market, albeit with localized constraints that distinguish it from the global release.
The Verge AI
© Matt WolfeModels & Labsmodels
Google Releases Gemini 4 Carbon for Cost-Efficient Inference
Google launched Gemini 4 Carbon, a new model variant aimed at providing high performance at a lower cost for inference tasks.
Matt Wolfe