Measured internally

Real numbers from our own hardware.

About one frame of latency on LAN — your input lands before the next frame draws. Steady 4K at 60 FPS. Hardware H.265, end to end. Every number on this page is measured internally with Remio's own stream-bench harness — no competitor comparisons, no third-party audit yet. The methodology at the bottom shows exactly how we tested, and how you can get a rough read yourself.

Latency

Speed you can feel.

Input-to-glass latency — the time between an input event and the moment the resulting frame renders on the client. Measured internally with our stream-bench harness on a LAN-equivalent path, with network conditions emulated in software rather than a live, variable network. Inside that frame, the processing pipeline itself — capture, hardware encode, the LAN hop, hardware decode — costs roughly 8 ms; the rest is the display's own 60 Hz refresh cadence, the same wait you'd have sitting at the machine.

LAN input-to-glass — measured internally, lower is better
One frame @ 60 Hz
16.7 ms
Windows host Fastest
~18 ms
Mac host
~22 ms

~17–23 ms

Imperceptible on a LAN. This is where Remio's Windows and Mac hosts measure internally — input generally lands before the next frame finishes drawing.

20–40 ms

Still smooth for productivity and most creative work. Typical for solid home Wi-Fi or a same-region relay hop.

40–80 ms

Workable, but you'll start to notice drag and scroll lag. Common on congested networks or longer-distance links.

80+ ms

Frustrating — visible delay on every action. Usually a rough network path, not the app.

Frame rate

Smooth all the way to 4K.

Pick the rate that fits your workload, from battery-saving to full 4K. Remio targets the rate you choose, and bitrate rides your network in real time so quality holds steady instead of stuttering.

30
fps
Battery saver
Ideal for productivity and remote work. Maximizes battery life on mobile clients.
60
fps
Balanced
The sweet spot for creative work, gaming, and everyday use — the ceiling every Remio stream targets today.
Auto
bitrate
Adaptive quality
Bitrate climbs when you have headroom and backs off before frames start dropping. Zero-buffering means you always see the latest frame, never a queued one.
Sustained frame rate — measured internally
4K No drop
60 fps
1080p
60 fps
Architecture

Built native. No shortcuts.

We haven't published CPU, memory, or install-size numbers yet — we want a measurement process we trust before we put figures on this page. What we can show you today is exactly how the pipeline is built.

Codec pipeline — hardware, not software
H.265 Primary
Hardware encode
H.264
Hardware fallback
Zero-buffering, by design
New frame arrives
Decoded & rendered now
Frame lost or late
Skipped, never replayed
Native runtime, every platform
iOS / macOS
SwiftUI
Android
Jetpack Compose
Windows
C++ / WinRT
Dynamic resolution
Resize the window
Virtual display follows, live
Go fullscreen
Matches your display exactly
Bandwidth

More pixels, sensible bits.

Adaptive encoding gets the most out of your connection. Five presets cover every scenario, all capped at a smooth 60 FPS — Remio shifts between them as your link changes.

Low
2 Mbps
720p
30 fps
Medium
5 Mbps
1080p
30 fps
High
10 Mbps
1080p
60 fps
Ultra
20 Mbps
4K
60 fps
Extreme
40 Mbps
4K
60 fps
Bandwidth ladder — same pipeline, five presets
Extreme Sharpest
40 Mbps
Ultra
20 Mbps
High
10 Mbps
Medium
5 Mbps
Low
2 Mbps
Startup time

From download to streaming, fast.

No accounts. No configuration wizards. No waiting room. Here's roughly what a first-time setup looks like, start to finish.

0 s
Download Remio
A small native installer — no bundled browser runtime to unpack.
~5 s
Install and launch
Native app installs instantly. No runtime dependencies to wait on.
~10 s
Pair with a code
Six-digit code on the host. Type it on the client, and rotate it anytime you like. No account required.
~15 s
Direct connection
Peer-to-peer by default. An encrypted relay only steps in if your network won't allow a direct path.
~20 s
Hardware encoder ready
Hardware-accelerated capture and encoding comes up on Apple and Windows hosts. The fast pipeline is live.
<30 s
Streaming
Full stream at the quality you picked, up to 4K at 60 FPS. Nothing left to configure.
Methodology

How we test — and how you can too.

Full transparency about what's behind every number on this page. We haven't published a public, reproducible benchmark suite yet — here's exactly what we did instead, and how you can get a rough read on your own.

Benchmark harness

  • Toolstream-bench (internal)
  • SetupHost + client, one machine
  • NetworkEmulated LAN & WAN profiles
  • PipelineReal capture → encode → decode

What we measure

  • MetricInput-to-glass latency
  • TimestampsIn-process, at send & render
  • Frame rateSustained fps under load
  • BitrateEncoder output vs. estimate

Hosts tested

  • WindowsHardware AMD AMF / NVENC
  • MacHardware VideoToolbox
  • CodecH.265 primary, H.264 fallback
  • VerificationInternal only, so far

These are in-house numbers. They come from our own stream-bench harness, run on a single machine with network conditions emulated in software — not a public, independently verified benchmark suite. That's on our roadmap. Curious about latency on your own setup? The simplest DIY check is two phone cameras: point one at the host screen and one at the client screen side by side, run a millisecond counter or stopwatch on both, and count the frame gap between moving the mouse and seeing it move. Or just download Remio and see how it feels.

Questions

Frequently asked questions

How does Remio measure LAN latency?
We measure input-to-glass latency internally, using our own stream-bench benchmark harness. A host process and a client process run the real capture, hardware encode, WebRTC transport, and decode pipeline, with network conditions emulated in software rather than a live camera rig. On a LAN-equivalent path, our Windows host measures roughly 18 milliseconds end to end, and our Mac host measures roughly 22 milliseconds — both close to a single frame at 60 Hz, about 16.7 milliseconds, meaning your input generally lands before the next frame finishes drawing. This is an internal measurement; we haven't published a public, independently verified benchmark suite yet.
Why don't you publish CPU, memory, or app-size numbers?
Because we haven't measured them rigorously yet. What we can tell you today: Remio is native code on every platform — SwiftUI on Apple platforms, Jetpack Compose on Android, and C++ with WinRT on Windows — with no browser runtime, no embedded web view, and no script bridge in the rendering path. Encoding and decoding go directly through the operating system's hardware codec. We'll publish resource-usage numbers once we have a measurement process we're confident in.
Can I reproduce these numbers myself?
Not with our exact tooling — stream-bench is an internal harness, not a public app. But you can get a rough read on latency yourself: install Remio on a host and a client on the same network, pair with the six-digit code shown on the host, and start streaming. Point two phone cameras at the host and client screens side by side, run a millisecond counter or stopwatch on each, and count the frame gap between moving the mouse and seeing it move on the other screen. We're working on a public, reproducible benchmark suite so these results can be checked independently — that's next on our roadmap.
Why does codec choice matter for remote desktop latency?
Codec choice sets a big part of the latency floor, and hardware encoders beat software ones by a wide margin — that's why Remio always prefers a hardware path over a software fallback. H.265 is Remio's primary codec end to end: hardware-accelerated on the host's encoder and the client's decoder. When a device can't do hardware H.265, Remio falls back to hardware H.264 instead of dropping to software encoding.
Are these numbers marketing claims or real measurements?
Measured — but internally, and not yet independently verified. Every latency and frame-rate figure on this page comes from our own stream-bench harness, run on our own hardware, with network conditions emulated in software rather than tested by an outside lab or a live camera rig. We're building a public, reproducible benchmark suite so these numbers can be checked independently; until then, treat this page as what we see in-house, not a certified result. Your own results will vary with your hardware, network, and codec support.
Remio app icon

Run the numbers on your own hardware.

Install Remio on the computer you want to reach and the device you reach it from. The methodology above lists every variable we used — see for yourself.