Apple did with Vision Pro what Apple does with a first-generation product it believes in: it left the industrial design alone and changed what was cheapest to change. The M5 refresh brings the current-generation chip, a revised strap that spreads load better, and nothing else you would notice across a room. That is a reasonable response to the two most common complaints — it felt slow under load, and it hurt after an hour.

What Apple did not change is the price or the shape of the proposition, and those are the reasons most people who tried one did not buy one.

What it is unambiguously good at

As a display, Vision Pro has no real peer. Text rendered at the centre of the field of view is sharp enough for sustained reading and code, which is the specific thing every previous headset failed at. A virtual monitor the size of a wall, in a hotel room or an open-plan office, is a genuine capability and not a demo.

Watching a film on it is the other easy win. A private cinema screen with good audio, at any seat on any aircraft, is a real experience rather than a novelty — for the roughly two hours the battery allows.

What the refresh does not fix

The device weighs what it weighs. A better strap redistributes that load; it does not remove it, and the front-heavy balance imposed by the glass front and the optics is inherent to the design. Sessions get longer with the refresh. They do not become all-day.

The battery pack remains external and wired. Apple traded head weight for a cable, which was the right call, but it means the device is never quite untethered and never quite furniture-free.

The software question is still the real one

Hardware this good makes the content gap more obvious, not less. There is a solid set of applications that run well, a smaller set built specifically for spatial computing, and a very small set you would buy the device in order to use. Three years into the category, that last set is what a $3,499 purchase depends on, and it is growing slowly.

This is not a criticism unique to Apple — every headset maker is in the same position — but Apple is the one charging flagship-workstation money.

Who should buy one

Buy it if you are building for the platform, if your work involves 3D or spatial content directly, or if a private large display in shared or mobile environments solves a problem you actually have. Those buyers get their money back in capability.

Everybody else is better served by a cheaper standalone headset for occasional immersive use, or by spending the same money on a very good laptop and a very good monitor. That is not a verdict on the engineering, which is genuinely impressive. It is a verdict on the price against what you can currently do with it.