Continuo sends your keyboard, mouse and clipboard between your own Macs over your own network. There is no cloud relay in the path: nothing is routed through a server, so what you feel is your LAN, your Wi-Fi radio, and the two Macs.
That also means your numbers are yours. Everything below comes from two specific Macs on one specific network, and each figure says how it was taken so you can judge it for yourself.
Measured
Four days of ordinary work, not a benchmark: 78 hours connected, with the pointer moving on the other Mac for about three and a half of them. Each row gives the typical figure and the worst it got.
| Typical | Worst | |
|---|---|---|
| Pointer round trip over Wi-Fi | 8 ms; 99% within 98 ms | 22 ms; 99% within 268 ms |
| Pointer round trip over the direct Mac-to-Mac radio | 38 ms; 99% within 269 ms | 161 ms; 99% within 535 ms |
| Held key, repeat spacing on the receiving Mac | 90% within 5 ms of the sending Mac | 99% within 22 ms; 2 of 777 arrived twice as far apart |
| Posting one scroll frame on the receiving Mac | 0.14 ms on average | 11 ms for the slowest frame |
| Disconnects while both Macs were awake | 7 in 78 connected hours | 176 s to reconnect, after a Wi-Fi channel change |
| File dragged to the other Mac | 10.7 MB/s (86 Mbit/s), 25 MB file | 8.6 MB/s (69 Mbit/s), 50 MB file |
For the two round trips, “worst” is the worst stretch of about 4,000 consecutive measurements, roughly 50 minutes, rather than a single unlucky moment.
Measured September 10 to 13, 2026, on a developer build of Continuo 26.9.9 beta that adds diagnostic logging. An M4 Pro Mac on macOS 15.7 drove an M1 Pro MacBook Pro on macOS 26.5, both on the same 5 GHz Wi-Fi 6 network on channel 44, with a signal around -62 dBm and no cable between them.
Day by day
| Day | Wi-Fi round trip, typical / worst 99% | Direct radio round trip, typical / worst 99% | Disconnects, both awake |
|---|---|---|---|
| September 10 | 8 ms / 160 ms | 37 ms / 326 ms | 1 |
| September 11 | 8 ms / 205 ms | 38 ms / 495 ms | 3 |
| September 12 | 10 ms / 268 ms | 121 ms / 532 ms | 0 |
| September 13 | 9 ms / 252 ms | 40 ms / 535 ms | 3 |
From late on September 11 until the early hours of September 13, both round trips got clearly worse and stayed that way. Signal strength stayed normal, and nothing inside Continuo was queuing, falling back or dropping input during it. That combination points below the app, at the radio or the network, and it is the reason the worst column is on this page at all.
What the numbers mean
Round trip on the pointer path. Continuo sends a small probe along the same UDP path the pointer uses, and the other Mac answers it. Timing that round trip uses one Mac’s clock at both ends, so it is a real duration rather than a comparison between two machines’ clocks.
The round trip is not symmetric: the probe goes out over UDP and the reply comes back over the encrypted control connection, so it is an upper bound. The one-way trip your pointer actually takes is shorter. The bound that can be defended is the one worth publishing; halving it and hoping is not.
The tail is also kinder than reality in one respect. A probe that never comes back leaves no measurement, and a reply that arrives after the next probe has gone out is timed against the newer one. The moments bad enough to break the connection are counted under disconnects instead.
Held keys. A held key repeats at a fixed rhythm, every 34 ms here. Comparing the spacing of the repeats as they left one Mac with their spacing as they arrived on the other measures jitter without comparing two clocks, because each spacing is timed on a single Mac. This covers 777 repeats from 69 held keys, on three of the four days.
Scroll. The time the receiving Mac spends handing each scroll frame to macOS. It is Continuo’s own share of the work at that end, and it is small next to a single 60 Hz display frame of 16.7 ms.
Disconnects. Continuo counts every time a connection goes down, including when a Mac goes to sleep. Sleep is not a fault, so every disconnect was checked against both Macs’ sleep and wake records, and against the MacBook Pro’s own macOS power log, which also catches a sleep that lasts only a few seconds. That left seven, all from the network:
- Six were the connection timing out, three of them in one bad stretch in the early hours of September 13. Two of those three took over a minute, because for that long neither Mac could reach the other at all.
- One was the MacBook Pro’s Wi-Fi moving to a different channel, which took 176 seconds to recover.
Half of the seven were back within 21 seconds.
File transfer. On September 18, two files were dragged from the M4 Pro to the M1 Pro across the screen edge. A 50 MB zip archive arrived at 8.6 MB/s and a 25 MB text file at 10.7 MB/s. Both were timed on the receiving Mac, from the moment the sending Mac announced the file to the moment the finished file was handed to the app it was dropped on. That window includes everything the drag did, not only the bytes in the air, so each figure is a floor: the network itself was not slower. These two drags ran on a later developer build than the rows above, with the M1 Pro on macOS 27 on the same channel 44 network. Two drags on one network are a first look, not a distribution. This row will be replaced once larger files have been timed several times.
What Continuo does not claim
There is no one-way latency figure on this page.
Measuring one honestly means comparing a moment on one Mac against a moment on another, and two Macs’ clocks are not the same clock. The difference between them is larger than the number being reported, so any one-way figure would be measuring clock drift as much as Continuo. The round trip above avoids the problem entirely by starting and finishing on the same machine.
Nor is there a headline transfer speed taken from the Quick Share mount. Reading a file out of that mount can exceed 600 MB/s, because at that point the file is already on your Mac and the read never touches the network. It is a real number and it measures the wrong thing.
How Continuo tries to stay fast
Pointer motion does not wait for retransmission. Cursor movement is sent as individual sealed datagrams. If one is lost the next one already carries a newer position, so a dropped packet costs you nothing. Waiting for a resend of a position that is already stale would cost you a visible stutter.
Two radios at once. Continuo sends motion over both your normal Wi-Fi connection and a direct Mac-to-Mac radio link at the same time, and uses whichever copy arrives first. The direct link is slower on its own, 38 ms against 8 ms above, because the Wi-Fi hardware has to share its time between your network and the direct link. It is not there to be faster. Either radio can stall on its own for a hundred milliseconds or more; both stalling at the same instant is far rarer, and that is the failure the second path exists to cover.
Keys and clicks are not droppable. They travel on a reliable, ordered connection, with a redundant copy over the fast path. A lost mouse movement resolves itself; a lost key release does not.
File transfers yield to the pointer. Bulk data is paced so that a large transfer in the background does not take the radio away from cursor movement while you are using it.
Checking your own network
The figures above come from a developer build that writes a detailed log. The version you download writes no such log, and Continuo sends no telemetry, so there is no Continuo figure on your Mac to compare against. What you can measure is the network underneath it, which is where most of the difference between two setups lives.
-
Hold Option and click the Wi-Fi icon in the menu bar. Note the channel and the RSSI, which is the signal strength. The run above was on channel 44 at around -62 dBm.
-
Find your router’s address in System Settings, under Wi-Fi, Details, TCP/IP.
-
On each Mac, while you use it normally, run this in Terminal with that address:
ping -c 150 -i 0.1 192.168.1.1 -
Read the last line. If the maximum regularly passes 50 ms, those pauses are in the radio or the network, below anything Continuo can see or fix, and they are what a stuttering pointer feels like.
What changes your result
Distance and walls, which band your Wi-Fi is on, how busy the network is, and what else the two Macs are doing. A 2.4 GHz network shared with a full household behaves very differently from a quiet 5 GHz one in the same room.
The channel matters more than it looks. Channels 44 and 149 are the ones the direct Mac-to-Mac link also uses, so a network on either spares the Wi-Fi hardware from hopping away to service it. Both Macs on Ethernet take the radio out of the path entirely.
If your Macs feel slower than the figures here, that gap is usually the network rather than Continuo. Connection and sleep covers what to check first.