Brave says its browser outperforms Chrome, Edge, and Firefox in real-world browsing conditions on macOS (Brave). The company reports 44% less CPU use on average than the three competitors, with the widest gap against Firefox: Firefox consumed 2.4 times as much CPU as Brave (Brave). The same report claims 28% lower memory use, 10% less energy per page load, pages loading 20% faster on average, and 26% less inbound plus 39% less outbound data transferred than the competing browsers (Brave).

Those are specific numbers, which is more than most vendor marketing bothers to offer. But the material available for review doesn't specify which macOS hardware or OS version was used, which browser versions were tested, or how many runs went into the averages. Without that, there's no way to know whether an outside party could reproduce the result.

Brave's figures describe what Brave measured, under conditions Brave controlled. What independent testing can and can't confirm about that picture is the more interesting question.

The clearest outside corroboration of Brave's efficiency claims comes from a MakeUseOf test that built a fresh 8GB Windows 11 virtual machine and opened 10 tabs across five browsers on their default settings (MakeUseOf). The tester measured "private working set" memory, the physical RAM actually claimed by each browser's processes, sampling every two seconds over a five-minute window after giving the pages two minutes to load (MakeUseOf). Each browser ran three times, with the median reported as the final figure.

The gap was wide. Brave's median RAM use came in at 738.8MB, against 1,460.7MB for Chrome, 1,827.4MB for Edge, 1,908.0MB for Opera, and 2,018.5MB for Firefox (MakeUseOf). That's roughly half of Chrome's footprint and about 37% of Firefox's. The tester flagged Chrome's result as a surprise: expecting a memory battle between Chrome and Edge, they instead found Chrome finishing second-lowest, using about 558MB less than Firefox, roughly 28% less RAM (MakeUseOf).

Brave also ran the fewest background processes by a wide margin, averaging 18.3 against a range of 47.7 to 61.7 for the other four browsers (MakeUseOf). Chrome, Edge, and Opera all share the same Chromium engine Brave is built on (MakeUseOf). If Brave and its closest engine relatives post such different memory numbers, the gap probably isn't coming from the engine itself.

The tester made that point directly: blocking behavior, tab-sleeping, and background process handling, not just the underlying engine, likely shaped these results (MakeUseOf). This is one data point, on one metric, on one platform, in one workload. It corroborates Brave's memory claim in the tested 10-tab Windows workload. It says nothing about CPU, energy, or bandwidth, and nothing about macOS, where Brave ran its own test.

Each row answers a different question. Private working set tells you how much RAM a browser is tying up right now, which matters most on a machine that's already short on memory. Brave's own figures describe what happens across an unspecified set of real-world browsing conditions, according to the company's description. A WebXPRT score measures how quickly a browser executes a fixed set of JavaScript-heavy tasks, independent of what a user happens to have open. A browser can lead on one of these and trail on another without any contradiction, because they aren't measuring the same behavior. Read as three separate, narrower questions, the results stop looking like a dispute.

The most plausible explanation for Brave's lower resource use is its built-in ad and tracker blocking, branded Shields. Blocking content can reduce the work a browser performs, since a script that never loads can't run, but the available tests don't measure how much of Brave's result comes from Shields specifically versus other factors. None of them compared Brave with Shields disabled against competitors running an equivalent ad blocker of their own, so the blocker's exact share of the advantage remains an inference, not a measured fact.

Brave has invested specifically in making that blocking cheaper to run. In January, the company rebuilt its Rust-based ad-block engine, moving roughly 100,000 default filters into a more compact FlatBuffers storage format and cutting the engine's own memory footprint by 75%, about 45MB by default across platforms (Brave). That's a documented improvement to the tool doing the blocking, separate from whatever savings the blocking itself produces by keeping ads and scripts from running in the first place.

History suggests this isn't settled science. A 2021 investigation found that Brave's older blocker code kept running even when disabled, adding roughly a second to page loads on the stable release compared with browsers running no ad blocker at all (Ctrl blog). The same tester measured the fix in detail: after Brave's developers addressed the issue within a few weeks, a preview build cut the time for the page-load event to fire by 81%, reduced blocking and unresponsive time by 95%, and dropped peak JavaScript heap use by nearly 73%, compared with the stable release still running the older code (Ctrl blog).

That episode shows Brave's blocker has previously added overhead rather than removed it, and that Brave fixed it once the problem was flagged. It also means no single benchmark, including this week's, should be treated as a permanent state of affairs. Whether Brave's current edge reflects a leaner browser engine or largely reflects doing less work by default is still an open question. Chrome or Firefox running a comparably strict ad blocker of their own could plausibly narrow some of that gap, but that's a testable possibility raised by the evidence, not a result any of the cited tests actually measured.

A WebXPRT 4 test from February 2025 scored Brave at 303, the lowest median among the browsers evaluated, with Firefox highest at 322, a difference of just over 6% (Principled Technologies). WebXPRT measures how quickly a browser executes a fixed set of JavaScript-heavy tasks, a different workload from the resource measurements above. The available research doesn't specify the platform or the full list of browsers scored in that round, so the result should be read narrowly: on this particular test, Brave finished at the bottom, not the top.

A 6% gap on a synthetic task suite is real, but it's not the kind of difference most people would notice while reading email or scrolling a news site. It matters more for anyone running JavaScript-heavy web apps, dashboards, in-browser editors, complex SaaS tools, where execution speed compounds across a session. For that use case, the evidence points away from Brave, not toward it.

The evidence supports a narrower, more useful answer than a single winner. What it actually shows depends on what a reader is optimizing for.

That last row matters. None of the cited research says anything about which browser handles extensions better or fits a particular workflow. This is a performance comparison, not a full buying guide, and it shouldn't be stretched to cover ground it doesn't touch.

The most useful missing test would be a controlled, cross-platform comparison that matches ad-blocking configuration across all four browsers and reports CPU, memory, energy, and load time together, rather than one metric at a time from three separate labs. Until that exists, the fair summary is narrower than "fastest browser": Brave has a credible, independently corroborated memory advantage in the tested 10-tab Windows workload, a plausible but unverified edge in CPU, energy, and bandwidth on macOS, and no demonstrated advantage on the cited WebXPRT score. That's the claim the current evidence actually supports, no more and no less.