# NetMaster: Network Tester — Speed, DNS & Dropout Tests

> Speed, DNS & Dropout Tests (iPhone/iPad app by Jack Richmond.)

- Source: https://appshunter.io/ios/app/netmaster-network-tester/id6794435820 (this page in markdown: same URL + `.md`)
- Developer: [Jack Richmond](https://appshunter.io/developer/1895145011)
- Category: Utilities, Developer Tools
- Price: Free
- Age rating: 4+
- Requires: iOS 17.0 · 7 MB
- Languages: American English
- Released: 2026-08-05
- Data updated: 2026-08-24
- User reviews in markdown: https://appshunter.io/ios/app/netmaster-network-tester/id6794435820/reviews.md

## What is NetMaster?

NetMaster is a privacy-conscious network diagnostic toolkit that helps you understand slowdowns, instability, and intermittent dropouts without an account, ads, analytics, or tracking.

RUN A COMPLETE NETWORK SCAN
• Measure download and upload goodput with parallel HTTPS requests
• See HTTPS latency, jitter, request failures, and a second latency vantage point
• See confidence ranges and a p95 only when enough observations support it
• Check direct UDP DNS response delivery across two providers, including bursts
• Check latency under load and receive a bufferbloat grade
• Compare everyday and first-time DNS response times
• See the DNS transport observed for NetMaster’s measured request and compare resolver performance
• Test real IPv4 and IPv6 reachability
• Time the router hop and see whether added delay is local or farther upstream
• Review public IP, ASN, egress region, TLS, HTTP, Wi-Fi security, captive-portal evidence, and tunnel indicators

NARROW DOWN CONNECTION PROBLEMS
Quick Diagnosis checks the path, router, internet reachability, DNS, and responsiveness in sequence, then gives one likely explanation and a practical next step.

INVESTIGATE INTERMITTENT DROPOUTS
Run the monitor for 5, 15, or 30 minutes. On iOS 26, a user-started finite session can request system-supported lock-screen continuation; earlier iOS versions must keep NetMaster visible. NetMaster records route changes, roughly one-second HTTPS heartbeats, gateway and DNS controls, IPv4/IPv6 evidence, latency changes, unknown coverage gaps, and optional access-point roaming evidence. Mark the moment you notice a dropout to compare your experience with the measurements.

UNDERSTAND WHAT WILL WORK
Plain-language readiness assessments explain whether the measured connection should handle video calls, 4K streaming, gaming, browsing, and large uploads. Protected local history makes it easier to compare the same network over time, with an optional metrics-only mode.

BUILT FOR HONEST RESULTS
A throughput-payload budget controls speed-test data use. Replenished parallel transfers target a stable measurement window; if a fast link exhausts the cap too soon, NetMaster labels the result as a lower bound instead of claiming it is the link maximum. HTTPS request timing is not ICMP ping. HTTPS failures and UDP DNS response loss are each labeled narrowly and are not presented as raw packet loss.

PRIVACY BY DESIGN
History stays in protected on-device storage and can be switched to metrics-only at any time. Dropout sessions use one redacted, protected recovery checkpoint so a crash or system expiration does not erase evidence. Reports leave the device only when you choose Share, with sensitive-address redaction enabled by default.

Important: Public iOS APIs do not expose exact Wi-Fi or cellular dBm, the connected Wi-Fi channel, channel width, or exact frequency band. NetMaster never invents those values. Wi-Fi identity is optional and requires system permission. A full scan transfers network test data; the Balanced default allows about 54.5 MB of throughput payload, plus small diagnostic requests and normal protocol overhead. Results are diagnostic evidence, not a guarantee or proof that a router, carrier, VPN, or service is defective.


## Version history (last 5 releases)

### 1.0.5 — 2026-08-22

Measuring satellite and aircraft connections properly, and never grading a connection on half the evidence.

• Upload speed on high-latency links. On an aircraft or a satellite connection, a round trip can take three quarters of a second, and the upload test's fixed time limit was being spent establishing connections rather than measuring one — so upload came back blank on exactly the connections people most want to check. The time limit now scales with the round trip the scan has already measured, the sizing probe shrinks on distant links, and transfers are sized to keep a long link genuinely busy rather than waiting on it.
• When a measurement cannot be taken, the report says which one and why. An empty field beside a working number is the least useful thing a diagnostic can print. Upload, first-hop timing, and the Wi-Fi standard now each explain their own absence.
• The latency-under-load grade no longer treats one direction as two. It has always graded the worst direction it measured, but a download-only grade looked identical to one measured both ways — and upload is usually the worse direction. The grade now says when it saw only one, and counts for less in the headline score.
• Fabricated DNS loss on slow connections is fixed. The direct UDP DNS check judged every request against the moment the run ended rather than its own waiting time, so on a link slower than about a second and a half it could report a burst of lost responses, a specific loss percentage, and a critical warning without a single response actually being lost. Requests are now judged individually, and a connection answering slower than the check waits is reported as unmeasurable rather than lossy.
• DNS advice that matches what was measured. Some networks degrade ordinary DNS while carrying encrypted DNS to the same providers perfectly — that is a network steering lookups to its own resolver, not a broken one. Where the scan sees exactly that, it now says so and points at encrypted DNS, because switching to a plain public resolver would send more of the traffic already being dropped.

### 1.0.4 — 2026-08-20

Multi-gigabit measurement, upload timing rebuilt, and a pass over every figure the app publishes.

• Multi-gigabit test profile: a 2 Gb or 2.5 Gb line — or Wi-Fi fast enough to carry one — spent the previous budget before the connection reached a steady rate, so it could only ever be reported as a floor rather than judged against your plan. The new profile measures to roughly 2.5 Gbps over a settled window, using a much larger payload budget and eight parallel connections instead of four. It moves well over a gigabyte on a link that quick, so it is a deliberate choice made on Wi-Fi rather than a routine scan.
• The profile previously called Maximum is now called Gigabit: the same budget, named for what it actually measures. Settings still states the fastest link each budget can measure before you scan, so you can pick one that can reach your plan.
• Upload speed is measured a different way. A phone reports a request body as sent the moment the system accepts it, so timing that moment measured how fast the socket took the data rather than how fast it left — which on one cellular connection produced 1941 Mbps beside a 200 Mbps download. Upload is now measured the way download always was, by sampling how many bytes have actually moved at two points inside a steady window.
• Your plan is compared only on the network you entered it for. A scan taken on airline, hotel, or guest Wi-Fi was previously measured against your home plan and reported as far below it, which is a category error rather than a finding. Settings shows which network the plan is judged on, with a control to move it.
• A sign-in page check that gives you the link. Hotel, airport, and aircraft networks hold you behind a portal, and the scan could report interception without offering any way to reach it. The Tests tab now checks in a couple of seconds and opens the page.
• An audit of every measurement the app publishes, and the fixes it produced. A download verdict is no longer suppressed by a short upload leg. Delay is no longer placed past the router while the first hop is bloating, because end-to-end delay contains the first hop and is now accounted for that way. The dropout report withholds a 95th percentile below twenty samples, and counts only time a probe actually succeeded as available rather than every second it watched. A failing DNS resolver now outranks "slow", which was reporting the symptom and hiding the cause.
• DNS privacy in shared reports: the encrypted-DNS availability result and the expected-resolver verdict now appear in the exported report as well as on screen.
• Classic DNS seen while a tunnel is active is reported as information rather than a warning, and says what the tunnel actually covered. A VPN encrypts your lookups as far as its exit; whether they stay encrypted past that point depends on the resolver at the far end, which no app on the device can see.

### 1.0.3 — 2026-08-14

Multi-gigabit measurement, and the DNS privacy findings now travel with your shared reports.

• Multi-gigabit test profile: a 2 Gb or 2.5 Gb line — or Wi-Fi fast enough to carry one — spent the previous budget before the connection reached a steady rate, so it could only ever be reported as a floor rather than judged against your plan. The new profile measures to roughly 2.5 Gbps over a settled window, using a much larger payload budget and eight parallel connections instead of four. It moves well over a gigabyte on a link that quick, so it is a deliberate choice made on Wi-Fi rather than a routine scan.
• The profile previously called Maximum is now called Gigabit: the same budget, named for what it actually measures. Settings still states the fastest link each budget can measure before you scan, so you can pick one that can reach your plan.
• DNS privacy in shared reports: the encrypted-DNS availability result and the expected-resolver verdict now appear in the exported report as well as on screen. The copy you send to a provider, or keep for your own records, carries the whole DNS picture instead of the transport line alone.
• Classic DNS seen while a tunnel is active is now reported as information rather than a warning, and says what the tunnel actually covered. A VPN encrypts your lookups as far as its exit; whether they stay encrypted past that point depends on the resolver at the far end, which no app on the device can see. The report says so plainly rather than implying nothing was protected.
• Measurement reliability on fast connections: test payloads are now sized from the link itself, staying large enough to be efficient on a quick connection while remaining small enough to finish on a slow one, so a result is not discarded for arriving too late.

### 1.0.2 — 2026-08-14

Bigger measurements on fast lines, plus new checks for DNS privacy.

• Maximum data budget: gigabit links finish a smaller budget too quickly to settle at a steady rate, so they could only ever be reported as a floor. The new budget measures them properly, and Settings now states the fastest link each budget can measure before you scan.
• Encrypted DNS: NetMaster checks whether this network permits DNS over HTTPS and DNS over TLS, and reports whether your own lookups actually travelled encrypted. Networks that block both are forcing your queries through a resolver they control.
• Expected resolver: name the resolver you intend to use, and any scan answered by a different one is flagged — the usual sign that a VPN or profile has quietly overridden your DNS settings while everything still looks encrypted.
• Fix & verify: record a change you made and NetMaster compares the scans either side of it, saying whether the difference is real or ordinary variation. It also settles whether Wi-Fi or the incoming line is the limit, and times the routes to services you actually use.
• A report you can send your provider, as text, CSV, or a formatted PDF.
• Sustained, replenished download and upload streams target a stable window while respecting the selected data cap.
• Adaptive latency and DNS sampling adds honest confidence ranges and withholds thin p95 figures.
• A two-provider UDP DNS delivery check reports missing, late, reordered, and burst responses without calling them generic packet loss.
• First-hop delay is measured separately during download and upload.
• On iOS 26, a user-started finite dropout test can request Apple's system-supported lock-screen continuation. Earlier versions remain foreground-only.
• History keeps up to 250 reports for time-of-day and cross-network comparisons.
• Layout and accessibility fixes, including large text sizes.

### 1.0.1 — 2026-08-11

Saved scans now keep their detail.

• History keeps the full report — Wi-Fi name, addresses, gateway and provider — instead of showing "Unavailable" for values the scan had just measured. Saved scans are listed by network name rather than a code.
• You can switch history back to metrics-only under More, On-device history. Turning it off also removes those values from scans already saved.
• Sharing is unchanged and still removes addresses by default.
• Download and upload are now measured against your plan separately, so a weak upload is no longer hidden behind a healthy download result.
• The Lock Screen and Dynamic Island activity for the Dropout Investigator is more dependable, and no longer keeps showing a session that has already ended.

## More apps by Jack Richmond

- [BizzBaits](https://appshunter.io/ios/app/bizzbaits/id6763063333)
- [DayBoard Companion](https://appshunter.io/ios/app/dayboard-companion/id6773437000)
- [Richmond Catch Crew](https://appshunter.io/ios/app/richmond-catch-crew/id6779360340)
- [Oyster Snob](https://appshunter.io/ios/app/oyster-snob/id6780932913)
- [LogGrade](https://appshunter.io/ios/app/loggrade/id6786964191)
- [Gadabout: Events Near You](https://appshunter.io/ios/app/gadabout-events-near-you/id6787334403)

All apps by Jack Richmond: https://appshunter.io/developer/1895145011

## Related topics

[network tester](https://appshunter.io/ios/topics/network-tester) · [btv network](https://appshunter.io/ios/topics/btv-network)

---

*Data collected daily from the US App Store and indexed by [AppsHunter](https://appshunter.io/). User reviews are verbatim App Store reviews. Ratings, prices and chart positions refresh continuously; this snapshot is from 2026-08-24.*