{"slug": "xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype", "title": "Xiaomi's new Xuanjie chips actually deliver on the local AI hype", "summary": "Xiaomi's new Xuanjie O100 chip, demonstrated in two engineering prototypes, delivers on-device AI performance with the foldable prototype achieving 300+ tokens per second (peak 330 tokens/s) running the Xiaomi MiMo 3B model offline, and the Xiaomi AI Cube running a 120B parameter model locally with 80 GB RAM (up to 160 GB). The prototypes showcase a tiered compute model that could reduce reliance on cloud-based LLMs, according to hands-on testing by XRING LAB.", "body_md": "# Xiaomi's new Xuanjie chips actually deliver on the local AI hype\n\nI managed to get hands-on with two engineering prototypes that demonstrate exactly how this silicon translates into a real-world AI workflow.\n\n## The Fan-Assisted Foldable: 300+ tokens per second on a phone\n\nThe first prototype is a foldable device powered by the Xuanjie O100. It looks like a standard foldable, but the back is a heavy-duty metal chassis featuring a circular intake/exhaust vent with an active cooling fan built right in.\n\nThe goal here wasn't just to make a phone, but to create a mobile AI powerhouse. Running the Xiaomi MiMo 3B model entirely offline (no Wi-Fi, no cellular), the responsiveness is staggering. Using the XRING LAB testing tool, I clocked a Time to First Token (TTFT) of roughly 0.45 seconds.\n\nThe generation speed is where it gets wild:\n\n**Average Generation Speed:**~303 tokens/s** Peak Generation Speed:**Up to 330 tokens/s\n\nAt these speeds, the AI is literally writing faster than you can read. For anyone interested in prompt engineering or real-time translation, this is the holy grail. You get instant, \"zero-latency\" interaction for document polishing or live transcription, all while maintaining absolute data privacy because the processing happens strictly on-device.\n\n## The AI Cube: Running a 120B model on your desk\n\nIf the foldable is about mobility, the \"Xiaomi AI Cube\" is about brute-force local compute. This thing is a CNC-machined piece of aerospace aluminum, designed to dissipate massive amounts of heat through 33,874 precision-cut cooling holes.\n\nThis isn't your typical mini-PC. It’s an Android-based AI terminal with 80 GB of RAM (supporting up to 160 GB) specifically built to deploy 120B parameter models locally. During the demo, we tested its coding capabilities:\n\n1. **Task Input:** A request to generate a functional piano web application.\n\n2. **Processing:** The system utilized a dual-model switching strategy (3B + 120B).\n\n3. **Output:** The 120B model handled the heavy lifting of logic and code structure.\n\n4. **Result:** Within moments, a fully functional \"Web Audio Online Piano\" was running on the screen, complete with real-time audio waveforms and interactive keys.\n\nThe \"small model + large model\" workflow is the clever part of this deployment. The 3B model handles high-frequency, lightweight commands instantly so the UI feels snappy, while the 120B model is summoned only when the task requires deep reasoning or complex coding. It effectively turns a desktop tool into a local LLM agent that doesn't need a cloud connection to function.\n\n## Hardware-driven AI differentiation\n\nSeeing these two prototypes side-by-side clarifies the future of the AI workflow. We are moving toward a tiered compute model:\n\n**Mobile/Foldable Tier:** Focused on high memory bandwidth (thanks to the O100's 1.22 TB/s bandwidth) to enable instant, offline interaction for daily tasks.**Desktop/Cube Tier:** Focused on massive parameter counts and sustained thermal performance to handle heavy-duty reasoning and development.\n\nThe bottleneck for local AI has always been the trade-off between model intelligence and latency. By optimizing the silicon specifically for these workloads, Xiaomi is showing that we might finally be able to move away from cloud-dependent LLMs toward a truly private, local AI ecosystem.\n\n[Apple is leaking its own hardware again 5d ago](/en/news/6881/)\n\n[Next Why AI-driven delusions follow a predictable spiral pattern →](/en/news/7592/)", "url": "https://wpnews.pro/news/xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype", "canonical_source": "https://promptcube3.com/en/news/7596/", "published_at": "2026-08-25 02:53:35+00:00", "updated_at": "2026-08-25 03:13:32.945920+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-chips", "ai-infrastructure", "ai-products"], "entities": ["Xiaomi", "Xuanjie O100", "Xiaomi MiMo 3B", "Xiaomi AI Cube", "XRING LAB"], "alternates": {"html": "https://wpnews.pro/news/xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype", "markdown": "https://wpnews.pro/news/xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype.md", "text": "https://wpnews.pro/news/xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype.txt", "jsonld": "https://wpnews.pro/news/xiaomi-s-new-xuanjie-chips-actually-deliver-on-the-local-ai-hype.jsonld"}}