Theo says a $700 Linux PC beats his $12,000 Mac
A T3 Code visualization comparing the two machines, with no published timings.
Theo posted a T3 Code visualization comparing the same work on two machines and noted the price gap in one line. "The Mac in this test cost $12,000. The PC cost $700." He called the visualizations his unexpected favorite feature ever.
The Mac in this test cost $12,000. The PC cost $700.
The complaint behind it is about spinning up threads. "Every time I spin up a thread on my Mac I am amazed at how much slower it is than my linux machines," he wrote, adding that he is thankful he moved all his code work to Linux.
Every time I spin up a thread on my Mac I am amazed at how much slower it is than my linux machines.
What is not in the post
No timings. Theo did not publish numbers in the post text, no model of Mac, no CPU, no spec for the $700 PC, and no description of what the test workload actually was. What exists publicly is a screenshot of a T3 Code visualization and a price comparison. Treat it as one person's impression of his own two machines, not a benchmark anyone else can rerun.
For working engineers
The reason this is worth a glance at all is that the thing being compared is process and thread spin-up, not raw single-core speed. Agent workloads look like a lot of concurrent processes being created, run and torn down, so the part of the machine that matters shifts toward how the operating system handles that churn. A box that loses on a single-threaded benchmark can still feel faster when you are fanning out subagents, and a box that wins on paper can still feel slow when every new thread costs you.
That is also the honest limit of the claim. Theo has the receipts for his own setup and nobody else has them yet. If your agent harness spawns heavily and your laptop feels sluggish in a way the spec sheet does not explain, this is a hypothesis worth timing on your own hardware rather than a result to repeat.
Have the good opinion, not the loud one
One email a day. Under three minutes. What working engineers found out about coding with AI.
