The Truffle is a miniature supercomputer designed to run multiple models (LLMs, machine translation, speech, etc.) without needing the internet. This device can pair with phones, monitors, and other devices to provide fast AI. No cloud needed! XRI Global has partnered with the developers of the Truffle to load our custom AI models for low-resource languages, especially those who don't have dependable internet or who can't afford the ongoing expense of Western cloud providers. If you would like to buy a Truffle and customize it for your use case and language(s), please get on the waitlist!
Multilingual Truffle Waiting List
Sign up to be notified when the product becomes available.
Full Name
*
First Name
Last Name
Company
Email
*
example@example.com
Language(s) requested
*
Please tell us a bit about your use case
Submit
Technical details about the Truffle
The Truffle has an NVIDIA Jetson AGX Orin. This is a high-performance AI computing module specifically designed for edge AI. It achieves 100+ TOPS, has an integrated CPU and GPU architecture for parallel processing, and has low power consumption relative to computational performance. For thermal management it has a custom-designed aluminum heatsink and an integrated cooling system. It comes with Bluetooth and a Wifi antennae. It can fit a 70 billion parameter model and get up to 22 tokens/second during inference.
Should be Empty: