Introduction to Local AI Equipment GuideBefore buying equipment, make sure it's worth going in, worth waiting for, and keeping.

Check the memory available for the actual model, not the display memory. It then shows the prefill for reading long inputs, the token generation rate followed by an answer, and the sustained load power and holding costs separately.

This site is for home users who are trying to narrow down what to buy between a finished Mac, an NVIDIA GPU, and a large integrated memory device. It's not an equipment store, it's a tool that provides comparisons and calculations.

Speed figures use collected benchmark measurements first. Only combinations without matching records are estimated from measured values. Pre-release hardware and distributed setups are labeled separately and updated after retail configurations and runtime support are confirmed.

You can review the calculation method, open data, privacy handling, and affiliate relationship separately. Hardware comparisons are calculated independently of affiliate-link availability.

For questions about the site or suggestions for data corrections, email us below.