Magnitude NetworkDePIN pilotMagnitude Radar: 14-day pilot report
From 29 October to 12 November 2026, 46 opt-in phones on Wi-Fi and mobile data checked the Magnitude RPC 1,026 times. 97.7% of checks succeeded and the average answer took 1.40 s. Most of that time is the distance between users and our server in the UK, not the node itself.
More phones and more checks every week
Daily checks grew from 9 on the first day to 108 on 11 November, from 7 phones to 41. By the second week each phone sent about 2.6 checks a day, out of at most 4. Android delays background work on phones that are rarely opened, so a check also runs when the app opens.
Checks per day
Phones that sent at least one check, per day
Full days in UTC, 29 October to 11 November. The first day started mid-day.
Response time from real phones
The average answer took 1,610 ms on 30 October and 1,343 ms on 10 November. Mobile data was a little slower than Wi-Fi.
Average response time per day
Wi-Fi and mobile data compared
| Network | Checks | Phones | Succeeded | Average | Fastest | Slowest |
|---|---|---|---|---|---|---|
| Wi-Fi | 550 | 30 | 98.4% | 1,336 | 324 | 5,000 |
| Mobile data | 475 | 34 | 97.1% | 1,480 | 416 | 4,628 |
| All networks | 1,026 | 46 | 97.7% | 1,403 | 324 | 5,000 |
Times in milliseconds, successful checks only. One more check came from another network type and failed; it is counted in the totals.
What Radar found: distance, not the node
The fastest check took 0.32 s, but the average was 1.40 s. A timing test on the server itself on 30 October shows where the time goes:
The node itself is fast. Most of the wait is the trip between testers' phones and our server in Manchester, UK, which is far from most of them. A monitor running next to the server would never see this. Next, we will serve the RPC closer to users, and Radar will measure the difference.
Failed checks
24 checks failed (2.3% of all checks): 21 timeouts and 3 network errors. No hour had more than one failed check, and other phones succeeded at the same time. That points to individual connections, such as weak signal or home Wi-Fi, rather than an outage of the RPC.
Who took part
116 people were invited and 47 switched Radar on (41%), with no rewards offered during the pilot.
What comes next
- Radar keeps running, and testers can see their own checks in the app (version 2.6.6).
- We will move the RPC closer to users and publish the before-and-after numbers.
- Rewards for Radar testers are being designed. Details will follow.
- We are opening Radar to partner chains and teams who want real-world uptime and speed for their own networks.
Have an Android phone and the Magnitude Network app? Join Radar and help measure the network from where you are.
Apply to join RadarHow these numbers were measured, and every day's figures
Each check is one JSON-RPC call (eth_blockNumber) from an opted-in Android phone to the Magnitude RPC, timed on the phone, with a limit of 5 seconds. A check succeeds when the RPC returns a block number. Figures cover testers who still had Radar on when the report was made, because turning Radar off deletes that phone's reports. Checks are kept for 30 days. No location or account details are part of this report.
| Day (UTC) | Checks | Phones | Checks per phone | Succeeded | Average (ms) |
|---|---|---|---|---|---|
| 29 Oct | 9 | 7 | 1.3 | 100.0% | 1,590 |
| 30 Oct | 25 | 12 | 2.1 | 100.0% | 1,610 |
| 31 Oct | 37 | 18 | 2.1 | 94.6% | 1,588 |
| 1 Nov | 46 | 19 | 2.4 | 97.8% | 1,524 |
| 2 Nov | 61 | 25 | 2.4 | 96.7% | 1,529 |
| 3 Nov | 70 | 32 | 2.2 | 98.6% | 1,485 |
| 4 Nov | 82 | 33 | 2.5 | 98.8% | 1,370 |
| 5 Nov | 87 | 32 | 2.7 | 97.7% | 1,437 |
| 6 Nov | 79 | 33 | 2.4 | 96.2% | 1,332 |
| 7 Nov | 83 | 33 | 2.5 | 97.6% | 1,310 |
| 8 Nov | 98 | 38 | 2.6 | 98.0% | 1,383 |
| 9 Nov | 104 | 39 | 2.7 | 98.1% | 1,346 |
| 10 Nov | 111 | 42 | 2.6 | 98.2% | 1,343 |
| 11 Nov | 108 | 41 | 2.6 | 97.2% | 1,381 |