Every way to type in Apple Vision Pro, compared
Typing is the unsolved part of spatial computing. The headset is remarkable at almost everything else and then asks you to enter a password, and suddenly you are poking at a floating pane with two fingers.
The five options
| Method | Good for | The catch |
|---|---|---|
| Look and pinch | Passwords, URLs, a search term | Fine for a few words, exhausting for a paragraph |
| Direct touch | Short bursts, feels natural | No surface, so no feedback — your fingers pass through the keys |
| Dictation | Long prose, alone in a room | Useless in an office, and punctuation is a chore |
| Bluetooth keyboard | Real work, real speed | You have to carry one, and it needs a surface anyway |
| Virtual keyboard on the table | Real work without carrying anything | Requires a flat surface in front of you |
Why direct touch is harder than it looks
Typing is not only a visual task. Touch typists rely on proprioception — knowing where the hands are without looking — and on the resistance of a key bottoming out. A floating keyboard removes both. There is nothing to stop the finger, so every keystroke has to be verified visually, and that verification is what makes it slow and tiring rather than the tracking accuracy.
The underrated factorArm fatigue. Holding hands up in mid-air for typing is the thing that ends a session, long before accuracy does. Anything that lets the hands rest on a surface changes how long you can work, not just how fast.
Realistic speeds
Individual results vary enormously, but the ordering is consistent:
- Look-and-pinch — fine for a dozen characters, impractical past a sentence.
- Direct touch — faster than pinch, tiring within minutes, error-prone without a surface.
- Dictation — the fastest input by far when the content suits it and you are alone.
- Physical keyboard — your normal typing speed, which nothing virtual currently matches.
- Virtual keyboard on a real table — between direct touch and physical, because the table supplies the missing feedback.
What actually changes the experience
Two things matter more than the input method itself.
A surface. Resting the hands on a desk removes fatigue and restores the stop at the end of each keystroke. That is why a virtual keyboard aligned to a real table behaves so differently from one floating in space, even though the tracking is identical.
Latency. Under roughly 50ms, typing feels connected to the hands. Above it, you start waiting for the screen to confirm what you did, and speed collapses regardless of how accurate the recognition is.
Pick by task, not by preference
- A password or a URL — look and pinch. It is fine, and it is right there.
- A message — dictation if you are alone, direct touch otherwise.
- An hour of real work — a surface, whether that is a Bluetooth keyboard or a virtual one on your desk.
The mistake is treating this as a single choice. Nobody uses one input method all day on a Mac either.