A U.S-based startup unveils the smallest AI Supercomputer called “Tiiny AI”, compact enough to fit into your pocket, but powerful enough to run 120 billion-parameter AI Models in your home.
Imagine holding the raw processing power of a data center in the palm of your hand. That is no longer science fiction, thanks to the newly unveiled Tiiny AI Pocket Lab. Developed by the US deep-tech startup Tiiny AI, this device has officially shattered expectations—and records—by becoming the world’s smallest personal AI supercomputer.
For tech enthusiasts, developers, and privacy advocates, the Tiiny AI Pocket Lab represents a massive shift in how we interact with artificial intelligence. No longer tethered to the cloud, this device brings “GPT-4o level” intelligence directly to your desktop, fully offline.
A Guinness World Record-Breaking Design
The Tiiny AI Pocket Lab isn’t just small; it is officially verified by Guinness World Records as “The Smallest MiniPC (100B LLM Locally)”. Measuring just 14.2 x 8 x 2.53 cm and weighing approximately 300g, it fits easily into a pocket. Despite its diminutive size, it is engineered to run massive Large Language Models (LLMs) ranging from the “golden zone” of 10B to 100B parameters, all the way up to 120 billion parameters.
To achieve server-grade performance in a handheld device, Tiiny AI utilized a combination of high-end hardware and breakthrough software optimization. The device features an ARMv9.2 12-core CPU and a custom heterogeneous module (SoC + dNPU) capable of delivering approximately 190 TOPS (Tera Operations Per Second) of AI compute power.
Perhaps most impressive for a device this size is its memory configuration. It comes equipped with 80GB of LPDDR5X RAM and a 1TB SSD. This massive memory pool is crucial for loading large AI models that typically require professional-grade GPUs costing thousands of dollars.
However, hardware is only half the story. The device is powered by two proprietary technologies: TurboSparse and PowerInfer. TurboSparse is a neuron-level sparse activation technique, while PowerInfer is an inference engine that dynamically distributes workloads between the CPU and NPU. Together, they allow the device to run heavy AI workloads efficiently without overheating.
Breaking the Cloud Dependency
For years, running a 120B parameter model meant renting expensive cloud servers. Tiiny AI aims to change that. As Samar Bhoj, GTM Director of Tiiny AI, stated, “Cloud AI has brought remarkable progress, but it also created dependency, vulnerability, and sustainability challenges”.
The Pocket Lab operates fully offline, meaning no internet connection is required to generate content, analyze data, or run complex reasoning tasks. This ensures bank-level encryption for user data, preferences, and documents, offering a level of privacy and persistence that cloud systems simply cannot provide. If you are a professional handling sensitive information, this device ensures your data never leaves your desk.
In an era where AI data centers are consuming vast amounts of electricity, the Tiiny AI Pocket Lab offers a sustainable alternative. It operates within a 65W power envelope, with a Thermal Design Power (TDP) of just 30W. This allows users to run high-performance models at a fraction of the energy and carbon footprint associated with traditional GPU-based systems.
A Versatile Ecosystem for Creators
The Pocket Lab is designed to be plug-and-play for developers, researchers, and students. It supports a one-click deployment ecosystem for dozens of leading open-source models, including:
- LLMs: OpenAI GPT-OSS, Llama, Qwen, DeepSeek, Mistral, and Phi.
- AI Agents: OpenManus, ComfyUI, Flowise, and SillyTavern.
Whether you need multi-step reasoning, deep context understanding, or complex agent workflows, this device brings the capability to your home without the lag or subscription costs of cloud APIs.
Availability
The Tiiny AI Pocket Lab was unveiled in December 2025, with a full feature showcase scheduled for CES in January 2026. It promises to democratize access to supercomputing, proving that the future of AI isn’t just in the cloud—it’s in your pocket.
To put this in perspective: relying on cloud AI is like having to drive to a massive public library every time you want to check a fact, paying for gas and adhering to their operating hours. The Tiiny AI Pocket Lab is like having the entire contents of that library instantly downloaded into your brain, available 24/7, wherever you are, without anyone else knowing what you are reading.
Pocket-sized AI Supercomputers in rise
The evolution of AI has led to a new class of small-sized supercomputers designed to bring the power of the data center directly to the developer’s desk, addressing concerns around cloud costs, latency, and data privacy.
NVIDIA’s DGX Spark, built on the Grace Blackwell Superchip, epitomizes this trend, delivering up to one petaFLOP of AI performance and 128GB of unified memory in a compact desktop form factor. This allows developers to fine-tune 70-billion-parameter models and run inference on models up to 200 billion parameters locally, bridging the gap between desktop development and cloud deployment.
Following that, the Tiiny AI Pocket Lab represents the extreme end of miniaturization, verified by Guinness World Records as the world’s smallest personal AI supercomputer; this makes high-performance, private AI accessible to developers and researchers without relying on the internet or high-end GPUs.
Key Takeaways
- The Tiiny AI Pocket Lab is the world’s smallest personal AI supercomputer, verified by Guinness World Records.
- It offers “GPT-4o level” intelligence directly to your desktop, fully offline, eliminating cloud dependency.
- The device features impressive hardware including an ARMv9.2 12-core CPU, 190 TOPS of AI compute, 80GB LPDDR5X RAM, and a 1TB SSD.
- Proprietary technologies like TurboSparse and PowerInfer enable efficient operation of large AI models.
- It supports a wide range of open-source LLMs and AI Agents for versatile use cases.
Join our community by subscribing to our Weekly Newsletter to stay updated on the latest AI updates and technologies, including the tips and how-to guides.
(Also, follow us on Instagram (@tid_technology) for more updates in your feed and our WhatsApp Channel to get daily news straight to your Messaging App).
