Upload Speed Test
Upload is the direction most plans neglect. This test sends real payloads to the measurement edge and reports the sustained rate, with every transfer listed.
Upload speed test
Real uploads to the nearest measurement edge, with every transfer listed.
Measured upload speed
— Mbps
Sampled from real transfers to the nearest measurement edge.
Peak sample
—Mbps
Data used
—
Duration
—
Show the individual transfers
| Transfer | Size | Time | Speed |
|---|---|---|---|
| No transfers yet. | |||
Each row is one real request. The reported result averages the last stable transfers and ignores the TCP ramp-up.
Ready. Nothing is measured until you start.
Why upload deserves its own test
Most consumer plans are asymmetric: a 300 Mbps download plan often includes only 10–30 Mbps of upload. That asymmetry is invisible when you browse and obvious the moment you join a video call in 4K, back up a photo library, or stream gameplay.
The upload test posts generated data to the measurement edge. Your browser reports upload progress events, which lets the tool build a timeline and exclude the buffering phase at the start of each request.
Where upload speed is the limit
| Activity | Comfortable upload | What you notice when it is short |
|---|---|---|
| Voice call | 0.5 Mbps | Robotic audio, dropped words. |
| HD video call | 1.5–3 Mbps | Your camera drops resolution or freezes. |
| Group HD call | 4 Mbps+ | Video turns off when others speak. |
| Live streaming 1080p | 6–10 Mbps | Dropped frames and buffering for viewers. |
| Cloud photo backup | 20 Mbps+ | Backups take days instead of hours. |
If upload is much lower than you expect, check whether the router is running in a legacy mode, whether a VPN is active, and whether something on your network is uploading in the background — cloud sync and security cameras are common culprits.
A note about upload measurement
Upload progress events are bursty: a fast link can report the whole payload in one or two events. The tool handles this by starting its timeline at zero and falling back to the full transfer when a grace window would leave no measurable remainder. Larger payloads are used for faster links so the measurement has enough duration to be meaningful.