Kneron, an on-device edge AI chip manufacturer, launched KL730, an auto-grade neural processing unit (NPU) chip. It will feature an integrated Image Signal Processor (ISP), as well as a peripheral interface that connects digital signals such as image, video, audio, and millimeter waves, to enable users to develop a number of AI applications across multiple industries.

The company claims that KL730 will offer a 3x–4x increase in energy efficiency compared to its previous models. Notably, the chip will also offer support for lightweight GPT large language models (LLMs), such as nanoGPT, and will be capable of 0.35–4 effective tera operations per second (TOPS).

 Furthermore, the chip will leverage Kneo, the company’s edge AI network, to enable users to run LLMs either partially or fully offline on their edge devices.