21 July, 2026

Low- and Mid-Tier Mobile for the Real World (2026)

Written by Harry Roberts on CSS Wizardry.

View this page as Markdown.

Table of Contents

Independent writing is brought to you via my wonderful Supporters.

1. What Makes a Representative Test Device?
2. DevTools Is an Approximation, Not an Answer
3. Real Low- and Mid-Tier Mobile Devices for 2026
   1. Samsung Again?
4. Real Low-Tier Mobile for 2026: Samsung Galaxy A17 5G
   1. Why Not the Galaxy A16 5G?
   2. The Galaxy A07 4G Complication
5. Real Mid-Tier Mobile for 2026: Samsung Galaxy A56 5G
   1. Why Not the Galaxy A57 5G?
6. What About WebPageTest Equivalents?
7. Closing Thoughts

Last year, I set out to find one representative low-tier and one representative mid-tier Android device for real-world web performance testing. Back then, I landed on the Samsung Galaxy A15 5G and Galaxy A54 5G. How much have things changed in a year?

Then, as it is now, the point wasn’t to find the slowest phones I could buy, nor was it to find two old flagships whose best days were behind them. Remember, low-tier does not mean old, and mid-tier does not mean a flagship from 2018. These are current device classes used by real people, with their own constraints around processing power, memory, storage, and price.

One year on, the method still holds up. Better yet, the market has given us much stronger evidence to work with. Let’s see what the best, real devices are for 2026…

tl;dr: If you want broadly representative Android hardware for real-world web performance testing in 2026, buy a Samsung Galaxy A17 5G (SM-A176x; 4 GB / 128 GB) for low-tier testing and a Samsung Galaxy A56 5G (SM-A566x; 8 GB) for mid-tier testing.

Both ranked among the world’s ten best-selling smartphones in Q1 2026, are widely ranged, receive long-term software support, and offer a meaningful separation in real browser performance¹.

And, if a large proportion of your audience is in the Middle East, Africa, or Latin America, I would also consider a Galaxy A07 4G as a regional, entry-tier class of device.

What Makes a Representative Test Device?

My original criteria considered four factors:

• Distribution and market penetration:
  • how likely is it that folk actually own the device?
• Device and hardware capabilities:
  • does the hardware authentically represent its intended tier?
• Support and longevity:
  • will the device remain relevant and secure for long enough to justify buying and maintaining it?
• Price-point:
  • is this a realistic consumer purchase at that tier?

For 2026, I’m adding a fifth consideration:

• Reproducibility and variant clarity:
  • can people in different markets buy comparable hardware, or does one model name conceal different internals?

This is to account for the fact that many manufacturers may vary device internals based on geographic locale and market, so a Galaxy A17 to you may well differ to a Galaxy A17 to me.

As ever, there is no single phone that can stand in for billions of Android devices. A broadly distributed default is useful in the general sense, but your own analytics should always get the final say. If most of your users are in one market, or you know that a particular handset dominates your traffic, test there too.

DevTools Is an Approximation, Not an Answer

Chrome DevTools’ calibrated CPU throttling is a useful starting point. It can approximate the CPU throughput of low- and mid-tier hardware relative to your development machine, while its network presets can constrain latency and bandwidth.

But DevTools now says this clearly: Throttling is relative to your computer’s capabilities […] DevTools can’t truly simulate the CPUs of mobile devices, because the architecture of mobile devices is very different from that of desktops and laptops².

Nor can a slowdown multiplier reproduce a phone’s available memory, storage performance, thermal behaviour, scheduler, background processes, or power-management decisions, so, again, you’re working through a very crude proxy. In fact, in three notable projects since last year’s piece, I found myself using—and getting markedly different results on—a physical device than I do in DevTools.

Real Low- and Mid-Tier Mobile Devices for 2026

	Low-tier	Mid-tier
Device	Galaxy A17 5G	Galaxy A56 5G
Model family	SM-A176x	SM-A566x
Recommended memory	4 GB	8 GB
Storage	128 GB; Notebookcheck’s tested unit used eMMC	128/256 GB; Notebookcheck’s tested 128 GB unit used UFS 3.1
Processor	Exynos 1330	Exynos 1580
Display	6.7″ AMOLED, 90 Hz	6.7″ AMOLED, up to 120 Hz
UK launch price	£199	£499 for consumer 8 GB / 256 GB
Software support	Six OS upgrades; six years of security updates	Up to six OS/One UI generations; six years of security updates
Q1 2026 global sales rank	Fifth	Seventh
Speedometer 3	7.07 in Chrome 140	12.0 in Chrome 134

The Speedometer figures come from separate Notebookcheck reviews, not a controlled head-to-head test. The different Chrome versions alone make a direct comparison difficult. Still, the A56 scoring roughly 70% higher is useful corroboration that these two phones occupy meaningfully different performance tiers³⁴.

Samsung Again?

At first glance, choosing two Samsung A-series phones for a second year running might look like lazy continuation, but I assure you, it isn’t!

I also considered current Redmi/Xiaomi, Motorola, Honor, OPPO/realme, Nothing/CMF, OnePlus, and Pixel devices. Several are excellent regional or price-specific alternatives—the Redmi A5 was the sole non-Samsung Android handset in Counterpoint’s Q1 2026 global top ten—but none combines model-level sales evidence, broad cross-regional availability, consistent hardware, long support, and the right performance constraints as neatly as these two.

Samsung’s grip on this particular part of the Android market is extraordinary: five of the six Android phones in that global top ten were Galaxy A-series devices¹, while the A56 was Europe’s best-selling smartphone of 2025⁵. The repetition is in a way a finding, and not just an assumption carried over from last year.

Real Low-Tier Mobile for 2026: Samsung Galaxy A17 5G

The Galaxy A17 5G is an unusually clean low-tier recommendation.

Samsung launched it in the UK in August 2025 at £199. The base configuration combines 4 GB RAM and 128 GB storage with an Exynos 1330 processor, a 6.7″ 90 Hz AMOLED display, six OS upgrades, and six years of security updates⁶. It subsequently launched nationwide in the US at $199.99, and official Samsung pages range the A17 5G in India, Australia, South Africa, and Singapore too⁷⁸.

That gives us the right kind of ubiquity: not jsut a very cheap phone that sells heavily in one region, but a current low-tier device available across a useful spread of markets.

More importantly, people are actually buying it. The Galaxy A17 5G ranked fifth among all smartphones globally in Q1 2026, making it the highest-ranked 5G Android handset in Counterpoint’s table¹.

The hardware gives us authentic constraints rather than novelty for novelty’s sake:

• Release date: August 2025 in the UK
• Cost at release: £199 UK; $199.99 US
• Chipset: Exynos 1330
• Recommended configuration: 4 GB RAM / 128 GB storage
• Screen: 6.7″ Super AMOLED, 90 Hz
• Support window: six OS upgrades and six years of security updates
• Why it’s representative: low launch price, broad international ranging, long support, and direct global sales evidence

Notebookcheck’s tested 128 GB unit used eMMC storage and scored 7.07 runs per minute in Speedometer 3 under Chrome 140. The reviewer also found that browsing could feel slow at times³. That is useful real-world feedback: the A17 is not simply a desktop CPU with a larger slowdown multiplier. Slow storage, limited memory, and lower main-thread headroom all combine to be tangible.

One thing worth noting: I would not assume that every regional A17 5G has the exact same storage technology just because Notebookcheck’s handset used eMMC. Samsung’s configurations vary by market, so verify the exact model you are buying. For a deliberately constrained test device, the 4 GB / 128 GB variant is the one I’d go for.

Why Not the Galaxy A16 5G?

The A16 has had much longer to build an installed base and remains a perfectly relevant phone to test if you already own one. But there is a difference between asking what do users have? and what should I buy now?

For a new purchase in 2026, I’d favour the A17. It occupies the same £199 launch position, is currently ranged, has a longer useful life ahead of it, and had already reached fifth place globally within months of launch.

The Galaxy A07 4G Complication

There is one solid argument against the A17: the Galaxy A07 4G ranked fourth in Q1 2026, making it the world’s best-selling Android handset in Counterpoint’s table¹.

Its sales contribution was particularly strong in the Middle East, Africa and Latin America. That makes the A07 arguably a more realistic floor for sites with a large audience in those regions.

Still, I would not replace the A17 with it as the single global default. The A07 4G is less far-ranged across the UK, Europe, and North America, while the A17 gives us a more reproducible cross-regional baseline. But this is precisely why one global recommendation will always be a compromise.

If your analytics lean heavily towards emerging markets, add the A07 4G as a regional entry-tier device below the A17. Your users are more important than broad averages, especially if they’re coming from emerging economies.

Real Mid-Tier Mobile for 2026: Samsung Galaxy A56 5G

The Galaxy A56 5G is an even easier recommendation.

Samsung launched the A56 in March 2025 with an Exynos 1580 processor, 8 GB RAM, a 6.7″ 120 Hz AMOLED display, and up to six generations of OS upgrades plus six years of security updates. The UK consumer model launched with 256 GB storage at £499; the US later received 128 GB and 256 GB variants from $499.99⁹¹⁰.

Ordinarily, choosing a device released in the previous calendar year might look a little conservative or cautious. Here, however, it is the evidence-backed choice:

• Omdia named the Galaxy A56 the top-selling smartphone model in Europe in 2025⁵.
• It then ranked seventh globally in Q1 2026¹.

That is about as direct an answer to how likely is it that folk actually have one of these? as we are going to get.

The hardware also provides a credible step above the A17 without drifting into flagship-lite territory:

• Release date: March 2025
• Cost at release: £499 UK for the consumer 8 GB / 256 GB model
• Chipset: Exynos 1580
• Memory: 8 GB
• Storage: 128/256 GB; Notebookcheck’s tested 128 GB unit used UFS 3.1
• Screen: 6.7″ Super AMOLED, up to 120 Hz
• Support window: up to six OS/One UI generations and six years of security updates
• Why it’s representative: mass-market reach, exceptional European sales evidence, global top-ten placement, and a material hardware step above the A17

Notebookcheck’s 128 GB test unit used UFS 3.1 storage and scored 12.0 runs per minute in Speedometer 3 under Chrome 134⁴. The faster processor, double the recommended RAM, and materially quicker storage give you a useful contrast with the A17—not just in raw benchmark throughput, but in cache access, memory pressure, and the amount of main-thread work the device can absorb before the experience deteriorates.

Why Not the Galaxy A57 5G?

Because newest is not the same as most representative.

The Galaxy A57 5G arrived in the UK on 10 April 2026, starting at £529 for the 8 GB / 256 GB model¹¹. As of 21 July 2026, that gives it only a little over three months of market history. It may well become the obvious mid-tier recommendation in time, but it does not yet have the A56’s installed-base or sales evidence.

This is absence of evidence, not evidence that the A57 is unpopular. If authoritative sales data later shows it displacing the A56, the recommendation should change. But picking it today would reward catalogue freshness over the very thing we are trying to measure: the hardware people actually use.

The A56 is not the recommendation despite being one generation old. It is the recommendation because the market has already proved it representative.

What About WebPageTest Equivalents?

WebPageTest’s open-source configuration does list its emulated devices. Each preset is a small configuration of viewport width and height, device pixel ratio, user agent, and a throttle_cpu multiplier¹². That gives us something a little more useful than inferring from the names shown in the device dropdown.

The closest presets I would use are:

Physical device	Closest WebPageTest preset	WebPageTest settings
Galaxy A17 5G	Google Pixel 5	393 × 722 CSS px; DPR 2.75; 2× CPU slowdown
Galaxy A56 5G	Google Pixel 6 or Pixel 7	412 × 786 or 412 × 783 CSS px; DPR 2.63; 1× CPU slowdown

The A17 sits between the Pixel 5 and Pixel 6 in general CPU benchmarks, but closer enough to the Pixel 5 to drop down a preset; the A56 is broadly Pixel 6/7-class. That is useful cross-referencing, but not a calibration: Geekbench is not a browser-based stress test, and WebPageTest’s slowdown is applied to its desktop test agent rather than modelling any real phone’s SoC¹³.

There is another limitation in the configuration itself. Pixel 6, Pixel 7, Galaxy S23, and every recent iPhone preset all use throttle_cpu=1: no additional slowdown at all. WebPageTest therefore collapses several materially different devices into the-performance-of-the-test-machine-they’re-running-on. Its agent can normalise CPU throttling across test machines, but it cannot make an unthrottled desktop behave like a particular phone¹⁴.

Further, the open source repository’s device file was last updated in December 2023, so names and values may change/have changed independently of this article. A physical A17 and A56 remain the references; the WebPageTest presets are convenient approximations when those phones are not available.

So, use these presets, but remember that these are the closest WebPageTest presets, not equivalent devices. (The network profile remains a separate choice and should be selected for the audience or demographic you’re testing against; it is not part of either device preset.)

Closing Thoughts

A representative test device is a solid reference point, but not always universal truth. The Android market is far too broad, regional, and fragmented for one phone to stand in for everyone.

But one well-chosen physical device is still vastly better than testing only on the powerful laptop on your desk and the flagship phone in your pocket. It exposes constraints that a slowdown multiplier cannot: memory pressure, slower storage, thermal behaviour, power management, and the point at which an interaction that felt fine on desktop turns into visible jank and annoyance.

For broadly representative Android testing in 2026, my choices are the Samsung Galaxy A17 5G (SM-A176x; 4 GB / 128 GB) for low-tier and the Samsung Galaxy A56 5G (SM-A566x; 8 GB) for mid-tier.

The A17 is current and already widely owned. The A56 proves why the newest model is not automatically the right one.

And if your audience is concentrated in emerging markets, the A07 4G is a useful reminder that the most honest baseline may be regional rather than global.

Happy testing!

---