Hardware Reviews

In-depth reviews of AI workstations, GPUs, mini PCs and homelab hardware, scored against a published methodology.

A review here is an assessment of what a piece of hardware can do for a specific kind of work — not a tour of its box contents.

Every review is scored against the same nine criteria, so that a 7 for thermals means the same thing on a mini PC as on a workstation GPU. The rubric, the score bands and what we measure are all published in our review methodology.

What every review tells you

  • The score, and its breakdown across all nine criteria — never a single number with no working shown
  • Who should buy it, and who should not
  • What it can actually run, stated with the quantisation, because “runs 70B” without qualification means nothing
  • The trade-offs, in plain terms — power draw, noise, memory that cannot be upgraded
  • What to consider instead, including the previous generation where it remains competitive

About our testing

We are a young publication and we have not physically tested most of the hardware we cover. Rather than imply otherwise, every review states its basis directly under the headline: specification analysis, third-party benchmark synthesis, or hands-on testing.

Specification-led analysis is genuinely useful — memory capacity, bandwidth, bus width and power draw are published facts, and a great deal follows from them. What it cannot tell you is how loud a machine gets after an hour at load, or whether the firmware is any good. Where those questions matter, we say that they are open.

Latest reviews

As an Amazon Associate, AI Gear Stack earns from qualifying purchases. Amazon and the Amazon logo are trademarks of Amazon.com, Inc. or its affiliates.