The ASUS UGen300 is a truly innovative AI accelerator, as it uses a compact and external format Connectable via the USB standard, bringing inference performance directly to any device.
The development of language models and artificial intelligence clients does not stop and is not only played in gigantic data centers and Cloud services. Computers for local development accumulate and reach the market in any type of format. The ASUS UGen300 is another novel example.
This AI accelerator is only slightly larger than a typical USB flash drive, with dimensions of 105 X 50 x 18 mm and incorporates the Hailo-10H AI processor, which offers 40 TOPS of dedicated power to support large language models such as LLM, VLM and others.
The UGen300 includes 8 GB of dedicated LPDDR4 memory and connects to other devices through an interface USB Type-Cconsuming as little as 2.5 watts of power under typical workloads. Its practical design plug-and-play Ensures cross-platform compatibility with Windows, Linux and Android operating systems. It is also compatible with major AI frameworks like TensorFlow, PyTorch and ONNX, ready to use.
“By integrating the Hailo-10H into a common USB device, ASUS puts the full potential of AI and generative AI within everyone’s reach”explains Max Glover, director of Hailo. “We are excited to see how our developer community will use this plug-and-play accelerator to push the boundaries of AI on devices. “This is precisely how Hailo envisions the future of AI: accessible, affordable, and designed so anyone can create with it.”.
Next-generation AI performance in a tiny size
The ASUS UGen300 USB AI Accelerator features a Hailo-10H AI processor that delivers dedicated 40 TOPS AI performance and is optimized for generative AI workloads, including LLM inference and vision-language tasks. Complements the CPU and NPUs of the host device and provides AI acceleration to free up system resources, enabling generative AI inference on the device without depending on cloud computing.
Compared to cloud-based AI, UGen300 users They pay no monthly subscription, experience zero latency, and enjoy great reliability and privacy.. UGen300 can run demanding next-generation AI applications such as text generation, video summarization, event triggering, speech-to-action, and real-time perception. Its low-power design uses just 2.5 watts to deliver efficient, high-performance AI at the edge of the network.
UGen300 is backed by a Optimized ecosystem designed to accelerate AI development. The upcoming UGen Utility tool will enable quick validation using over 100 pre-trained models for an easy start. Users can also connect to the Hailo developer community to access tutorials, reference designs, and shared information that will help them build and deploy AI applications at the edge faster.
