iPhone 18 Pro Max review of the A20 Pro: Farooq's benchmarks, low power mode, and the twice-as-fast Draw Things result explained. Read the numbers first.
iPhone 18 Pro Max review: what the measured numbers show
This iPhone 18 Pro Max review is built on Farooq's measured runs and the benchmarks Apple publishes on its own iPhone 18 Pro page, which is the primary source for every vendor figure below. The phone's gains are real but uneven: raw CPU and GPU tests reproduce or beat Apple's headline numbers, while compute, thermal, and battery claims need tighter reading. The A20 Pro is Apple's 2026 flagship system-on-chip, built on the third-generation 2-nanometer process.
Farooq is the host of the YouTube channel iPhonedo, which published this review on 16 September 2026. His caption states absolute results on Geekbench 6 and Geekbench 5 and compares them with Apple's own published claims rather than treating them as interchangeable.
The chips in question are the Apple A20 Pro in the iPhone 18 Pro and Pro Max and the A19 Pro in the iPhone 17 Pro, Apple's 2025 flagship phones. The iPhone 18 Pro Max is the larger 2026 model; the iPhone Air is Apple's 2025 ultra-thin phone.
A20 Pro benchmarks vs Apple's claims
The A20 Pro's CPU gains clear Apple's targets, its GPU peak slightly misses, and its compute gain runs far ahead of Apple's footnote. The table below puts Farooq's absolute scores beside Apple's published relative claims so the two evidence classes stay separate.
| Test | iPhone 18 Pro Max (measured) | vs iPhone 17 Pro Max | Apple's published claim |
|---|---|---|---|
| Geekbench 6 single-core | 2,631 | +24% | Up to 25% faster CPU |
| Geekbench 6 multi-core | 7,602 | +22% | Up to 25% faster CPU |
| Geekbench 6 GPU peak (Metal) | 53,351 | +29% | Up to 46% faster graphics |
| Geekbench 5 Compute (Metal) | 44,113 | +45.5% (vs 30,318) | Up to 60% faster Neural Engine compute |
| Geekbench 5 Compute, Low Power Mode | 33,745 | +27.5% (vs 26,466) | Not published |
The chip has a six-core CPU, a six-core GPU, a 16-core Neural Engine, and hardware ray tracing, matching the A20 Pro listed in Apple's iPhone 18 Pro specifications. Apple's performance claims also appear in the iPhone 18 Pro press release published at launch.
Farooq also cites a separate "compute score" figure of 45% over the iPhone 17 Pro Max that matches neither a Geekbench 5 nor a Geekbench 6 Metal run under its own maths. Treat that number as unresolved rather than combining it with either run.
Low power mode and the thermal redesign
Low power mode narrows the A20 Pro's advantage rather than erasing it. Farooq's Geekbench 5 Compute run drops from 44,113 to 33,745, a 23.5% fall on the new phone, while the A19 Pro in the iPhone 17 Pro Max falls from 30,318 to 26,466, a 12.7% fall. The 18 Pro Max keeps a 27.5% lead in low power mode, down from 45.5% at full performance.
Those figures are Farooq's own Geekbench 5 runs as captioned in the review, not Apple measurements. Geekbench 5 and Geekbench 6 are different benchmark versions with different scoring, which is why the 24% single-core figure appears only under Geekbench 6 and the 45.5% figure only under Geekbench 5.
The thermal design is the structural change. The 2026 phones pair the A20 Pro with a larger vapor chamber than the iPhone 17 Pro used, and the chip separates the silicon die from the unified memory instead of stacking them. Farooq's thermal-camera tests show the new phones dissipate heat over a wider area and cool down faster than the 17 Pro Max, though the iPhone Air still cooled fastest in his comparison.
Practical effect: the 18 Pro Max sustains demanding tasks such as gaming for longer before clocking down. Farooq measured this with a thermal camera, so it is his own observation method rather than an instrumented independent test.
On-device AI: what reproduced and what did not
One AI figure from this iPhone 18 Pro Max review does not hold up. Farooq attributes a 25.4% on-device AI gain to a "synthetic benchmark", but that number does not match his own Geekbench runs and he does not name the tool, so it cannot be traced.
The test that can be traced is Draw Things, an on-device image-generation app that runs diffusion models locally. Farooq reports the iPhone 18 Pro Max finishing the same generation task almost twice as fast as the comparison phone, and the transcript's comparison device needs checking before the figure is quoted.
Apple's own claim for the Neural Engine is up to 60% faster compute, reported on its iPhone 18 Pro page and in the launch newsroom release. The Draw Things result is a different, broader workload that chains several compute stages, so it is not a direct measurement of Neural Engine throughput.
Realistic takeaway: image generation and on-device editing tasks benefit from the A20 Pro, but the exact multiple depends on the model, prompt, and app version. Only the workload name and the relative outcome are worth repeating.
Camera: variable aperture, Pro Settings, and where they stop
The main camera's aperture now moves between F1.48 and F4 where the iPhone 17 Pro was fixed at F1.78. Farooq's numbers: at F1.48 the sensor takes in roughly 44% more light than F1.78, and closing to F4 cuts about 86% of that light, because each stop halves the area. The narrower end gives brighter-day footage less need for extreme shutter speeds and produces a starburst effect around highlights.
Pro Settings adds manual aperture, white balance from 3,500K to 7,500K, manual focus with peaking, manual shutter speed, and a Lovable histogram, plus 4K time-lapse. Manual ISO and manual focus are still missing in photo and video modes, so tools like the FiLMiC Pro camera app remain necessary for full manual control.
Cinematic mode now records at 60 frames per second with slow-down in post, and footage shot in ordinary video mode on the main camera can be converted to cinematic afterwards. Focus tracking held a playing card even when it left the frame and the deal accelerated, which Farooq demonstrated on camera.
Editing gains include texture controls and grain in photo Styles, plus a Clean Up tool that falls back to Apple's foundation models when its on-device attempt fails. Apple's reference-image feature stores a main-camera original and can show that a photo has not been altered.
Battery, modem, and market split
US models use a Qualcomm Snapdragon modem while models sold elsewhere use Apple's C2 modem, a split Farooq notes because it prevents him from testing C2 directly in the US. Apple does not publish battery capacity in watt-hours on the specifications page, and no reviewer has produced independent runtime results yet, so capacity claims should be avoided.
Both phones gained battery capacity over the iPhone 17 Pro, and 45W fast charging is confirmed on Apple's iPhone 18 Pro specifications page. The combination of a larger battery and the 23.5% low-power-mode reduction Farooq measured on the new chip is the strongest reason his battery claim is credible.
What is not confirmed: any specific hour count. Without an Apple rating or an independent rundown, hours cannot be estimated from a chip benchmark. The chip's low-power behaviour is only one input into screen-on time, and the 23.5% reduction in compute throughput is a performance trade, not an efficiency measurement.
The practical reading is that this is a "no numbers yet" claim, not a weak one. The parts that can be verified are the modem split and the confirmed charging figure; the runtime claim needs a separate test.
Smaller Dynamic Island and why old cases do not fit
The Dynamic Island is now smaller, and Apple achieved it by lowering pixel density on that part of the display so the infrared camera can capture the face for Face ID through it. Under a microscope the arrangement is visible, but Farooq reports no practical difference at normal viewing distance.
The cutout now supports up to three simultaneous Live Activities, which can be rearranged with a swipe. That is a software behaviour tied to the current iOS version, and iOS 27 is the major release associated with this generation.
Case incompatibility comes down to the camera plateau. By Farooq's calipers the plateau grew from 4.3mm to 4.5mm and individual lens heights from 1.8mm to 2.1mm, matching Apple's iPhone 18 Pro specifications. Because the camera module protrudes further, iPhone 17 Pro cases do not fit iPhone 18 Pro phones, which is why Apple's own case listing states incompatibility.
Farooq's personal recommendation from first-hand use is the new wrist strap over the previous body strap, though that is a user preference rather than a measured result.
Frequently asked questions
- Is the iPhone 18 Pro Max faster than the iPhone 17 Pro Max? Yes. Farooq's Geekbench 6 runs show 2,631 single-core and 7,602 multi-core on the iPhone 18 Pro Max, both roughly 22 to 24% above the iPhone 17 Pro Max. Apple's own page claims up to 25% faster CPU, so the two evidence classes agree.
- Does the iPhone 18 Pro Max have a variable aperture camera? Yes. The main camera moves between F1.48 and F4, while the iPhone 17 Pro was fixed at F1.78. Farooq measured roughly 44% more light at the widest setting and about 86% less light at the narrowest compared with the older fixed lens.
- Is low power mode slower on the iPhone 18 Pro Max? Yes, and by more than on the previous model. Farooq's Geekbench 5 Compute run dropped from 44,113 to 33,745 under low power mode, a 23.5% reduction, while the iPhone 17 Pro Max dropped 12.7% on the same test.
- Which modem is in the iPhone 18 Pro Max? US models use a Qualcomm Snapdragon modem, and models sold elsewhere use Apple's C2 modem, according to Farooq. That split means US reviewers cannot test the C2 directly on the current generation.
- Can I use iPhone 17 Pro cases on the iPhone 18 Pro Max? No. Apple's own case listings state incompatibility, and Farooq's calipers explain why: the camera plateau grew from 4.3mm to 4.5mm and the lens heights from 1.8mm to 2.1mm.
How Farooq tested, and what that means for these numbers
The benchmark and thermal results in this review come from Farooq's own test setup, and the method matters as much as the scores. iPhonedo is the YouTube channel that published the review on 16 September 2026, so every figure attributed to Farooq is a first-hand observation from that video rather than an independently reproduced measurement.
His equipment included a thermal camera for the vapor-chamber comparisons and the Draw Things app for image generation. He does not state ambient temperature, device skin temperature, or the firmware and app build used for each run, which limits direct reproducibility.
Three separate evidence classes run through this article. Apple's published claims on its iPhone 18 Pro page are vendor figures. Farooq's Geekbench and Draw Things results are his own measurements. The 25.4% AI figure he cites from an unnamed "synthetic benchmark" matches neither, and its attribution remains unverified.
Turn a video review into a written article
A review like this one lives on measurements, comparisons, and the way a host explains a feature. Anyone who has recorded similar material on a phone launch, a developer tool, or an interview already owns the raw content a written article needs.
Skalablog takes a YouTube URL, produces a transcript, and generates a structured article from it. Paste a video into Skala blog and the spoken review becomes something a reader can scan, quote, and search.
Fork this article
Start a new branch from the same video, shaped your way. You keep the credit; the original keeps the attribution.
A fork in another language is filed as a translation of this article, so the two pages point at each other. You can unlink it later from the editor.
0/240
You are creating
- Format
- For
- Language
- Source
- Your angle
You will be asked to sign in before it is generated.
Buy credits