By Ecosystem Analyst | 27 September 2026 | Pillar Research
Ecosystem Analyst, Apple Unboxed. I recomputed published battery sheets. I did not discharge a handset for this piece, and I did not survey 50 owners.
How this was tested: It was not tested in a lab I control. On 27 September 2026 I opened Apple’s spec sheet, three independent write-ups, and one five-day diary, converted every printed time to minutes, and recomputed the gaps. The queue for this week asked for a 50-person, six-month survey. I cannot field that from a desk, and I will not invent it.
Key takeaway: Apple’s iPhone 17 Pro battery life claim and the lab rundowns are different tests wearing the same noun. On the sheets I could actually recompute, the eSIM cell is 254 mAh larger than the nano-SIM cell (6.35 percent) and GSMArena’s Active Use Score moved 49 minutes (5.31 percent). Their video rundown moved the wrong way, by -7 minutes. If you buy the phone because a keynote hour and a review hour are the same hour, you are buying a sentence, not a result.
What I set out to measure, and what I refused to fake
The working title in the queue was “iPhone 17 Pro Battery: 6-Month Real-World Data From 50 Users.” The method on that card was a survey plus a personal log. A weekly cron does not have 50 volunteers, six months of cycle counts, or a drawer of 17 Pros. Publishing that title with made-up percentages would be the opposite of the reason this series exists. AdSense and a careful reader both punish a research costume. I kept the keyword, iPhone 17 Pro battery life, and changed the method.
What I did instead is Method B. I aggregated public primary numbers and did the arithmetic myself. The original data in this post is the recomputed table: minute conversions, percentage gaps, milliamp-hour per hour of GSMArena’s mix, and a straight-line reading of Apple’s 1,000-cycle endpoint. The raw times belong to the labs. The gaps, and the decision to keep unlike tests in separate columns, are mine.
If you already own the phone and want the settings that move a real day, start with the settings that actually change iPhone battery life. This piece will not tell you to turn Background App Refresh off. It will tell you which published number is a claim, which is a rundown, and which is a diary that does not reproduce its own summary.
Research methodology
I worked on 27 September 2026. The phone had been on sale for about a year. That is long enough for launch reviews, a second-wave eSIM retest, and at least one published multi-day diary to exist. It is not long enough for me to have run a six-month panel, and I did not pretend otherwise.
Inclusion rule
A number enters the dataset only if I could read it on a page I opened in this session, or on a page whose table I re-fetched and checked against the first extract. Apple’s iPhone 17 Pro spec sheet supplied the official video hours, the charge claims, the eSIM sentence, and footnotes 7 and 8. Apple Support 106348 supplied the cycle rating, and I confirmed the sentence in the fetched HTML, not from memory: batteries of iPhone 15 models and later are designed to retain 80 percent of original capacity at 1,000 complete charge cycles under ideal conditions. iPhone 14 and earlier stay on the 500-cycle line.
GSMArena’s 30 September 2025 review and 10 November 2025 eSIM retest supplied both cells, the four component rundowns, and the charge curve they measured without Apple’s own 40W brick. Tom’s Guide’s 17 Pro review was re-fetched as HTML. On that page the drain test is 15 hours 32 minutes, the 16 Pro comparison is 14 hours 7 minutes, and the charge table is 40 percent at 15 minutes and 72 percent at 30 minutes on Apple’s 45W USB-C charger. The same page prints battery size as N/A. I did not borrow a milliamp-hour figure from a Tom’s Guide comparison URL that returned 403 to the extractor.
Trusted Reviews’ 18 December 2025 review supplied the UK physical-SIM unit, the 3,998 mAh row, the Netflix and gaming spot checks, and the only 20-minute charge mark I could find (51 percent, against Apple’s 50 percent claim). Victor Hristov’s 30 January 2026 PhoneArena diary supplied five drain-and-screen-time rows, 52 cycles, and 100 percent health on a global unit. 9to5Mac supplied The Tech Chap’s YouTube times that the article actually printed: 7 hours 29 minutes for a 16 Pro Max, 7 hours 58 minutes for a 17 Pro Max, 6 hours 43 minutes for the Air, 6 hours 55 minutes for the base 17. The 17 Pro “beat the 16 Pro Max” in that write-up, but no minute count for the 17 Pro was printed. I did not invent one from the phrase “about an hour and a half.”
MacRumors’ 10 September 2024 battery comparison supplied the iPhone 16 Pro baseline I could still open: 27 hours of video playback and 22 hours of streaming video, 33 and 29 on the 16 Pro Max. Apple’s live iPhone 16 Pro URL redirected to the current iPhone page when I requested it, so those 16-series hours are a launch-week transcription I re-read, not a spec sheet I could reopen on Apple.com today.
Exclusion rule
I opened an Apple Community thread that search snippets described as a 17 Pro at 90 percent after 329 cycles. The live page was a bot wall, not the post. That number is not in the dataset. YouTube comments under year-later videos are not a sample I collected, so they are not in the dataset either. A secondary site that said it had scraped about 1,500 Reddit posts is their scrape, not mine. I did not launder it into a fake panel of 50.
I also excluded any figure that required me to treat two protocols as one. Apple’s video-playback test, GSMArena’s video rundown, Tom’s Guide’s web-surfing drain, Trusted Reviews’ one-hour Netflix spot check, and The Tech Chap’s continuous YouTube session are five procedures. Putting them in one column and calling the difference “the truth about battery life” would be a rhetorical trick. They sit in separate columns. Where I show them on one chart, the caption says the chart is a warning label.
Conversion and rounding
Every hh:mm value became minutes before any percentage was taken. 15:32 is 932 minutes. 14:07 is 847 minutes. The year-over-year Tom’s Guide gap is 85 minutes, which is 10.04 percent, not “about 10 percent” unless I show the rounding. Percentages in the prose are printed to two decimal places when the gap is the finding, and I note where a lab rounded more coarsely. GSMArena called the call-test gain 3 percent. The minute counts they printed, 24:01 against 23:14, are 47 minutes and 3.37 percent. Both statements can stand. I am not correcting their headline. I am showing the arithmetic their table supports.
Capacity percentages use the prose figure the lab led with, then the conflicting label if one exists. GSMArena’s review prose says 3,998 mAh for the physical-SIM cell. Their charging chart on the same review labels that phone 3,988 mAh. PhoneArena’s global-model cell is 3,988 mAh. I did not average 3,988 and 3,998 into 3,993. A 10 mAh smear would hide a labeling conflict that a buyer cannot resolve from the outside.
The straight-line cycle reading
Apple’s endpoint is 80 percent at 1,000 cycles under ideal conditions, and the same paragraph says the exact percentage depends on how the phone is used and charged. A straight line from 100 percent at cycle 0 to 80 percent at cycle 1,000 drops one point per 50 cycles. Apple did not publish that line. I drew it so a reader can see what “on pace” would mean if wear were even. It is a ruler, not a forecast. Heat, charge limits, and overnight 100 percent sits will move a real pack off that ruler. The battery-health guide on this site covers what the Settings screen is and is not saying. This methodology does not repeat that guide.
Controls I did not have
I did not set brightness, I did not choose the cellular band, and I did not hold the phones. GSMArena describes a repeatable mix of calls, web, video, and gaming. Tom’s Guide, on a separate methodology page I did not need to treat as a new data row, has long described a web-surfing drain at a fixed brightness over cellular. Apple’s footnote 8, on the spec sheet I opened, says the charge test used preproduction units in July 2025, an Apple 40W Dynamic Power Adapter (model A3351), and timing from the appearance of the Apple logo. Trusted Reviews and Tom’s Guide used retail phones and named different bricks. Those are not failed replications of Apple’s protocol. They are different protocols that happen to share a 15-minute mark.
Sample size, stated plainly: four published lab write-ups for the 17 Pro, one five-day diary from a named reviewer, one YouTube shootout whose printed 17 Pro Max times I used only as a contrast, and Apple’s own claim sheet. That is not 50 users. Anyone who cites this post as a 50-person study is misreading it.
What I counted as a finding
A finding had to survive a recompute. If my minutes matched the lab’s own summary, I treated that as a check, not as news. The news is where the summaries diverge from each other, or where a summary does not reproduce from the table under it. GSMArena’s 49-minute Active Use gap matches 49 minutes. That check passed. PhoneArena’s “5 hours and 20 minutes per 100 percent” does not fall out of the five rows in the same article. That check failed, and it is in the findings for that reason.
Apple’s hours are a claim with a date and a footnote
On the spec sheet, iPhone 17 Pro video playback is up to 33 hours. Streamed video playback is up to 30 hours. The Pro Max is up to 39 and 35. Footnote 7 says all battery claims depend on network configuration and many other factors, actual results will vary, and the battery has limited recharge cycles. Footnote 8 dates the fast-charge test to July 2025, preproduction units, and a specific Apple adapter. The product page repeats the 33 and 39 hour local-playback figures and calls the Pro Max battery life the best ever in an iPhone, with a comparison back to the 15 Pro Max rather than a methods section.
Against the 16 Pro transcription I could still open, local playback moved from 27 hours to 33, which is 6 hours and 22.22 percent. Streaming moved from 22 hours to 30, which is 8 hours and 36.36 percent. The Pro Max local figure moved from 33 to 39, also 6 hours, which is 18.18 percent because the base was higher. Streaming on the Max moved from 29 to 35, 20.69 percent. Same absolute hour gain, different percentage, different test than any review rundown.
I am not saying Apple measured those hours incorrectly. I am saying the sentence “up to 33 hours” is the output of Apple’s video-playback procedure on the units and software in that footnote. It is not Tom’s Guide’s web test, and it is not a promise that your Thursday will look like a spec table. The six-month Pro Max review on this site is a different kind of piece, a use report rather than a sheet recomputation. Read them as neighbors, not as duplicates.
There is a documentation split inside Apple’s own support pages that I am leaving visible. Support 106348, fetched today, applies the 1,000-cycle / 80 percent design target to iPhone 15 models and later. Support 101575, which I also opened, still phrases the 1,000-cycle line as “iPhone 15 models” in the paragraph I extracted, without the words “and later.” I am not going to decide which page a lawyer would prefer. For a 17 Pro owner, 106348 is the page that explicitly includes later models. The practical instruction on both pages is the same: the percentage you see depends on use, and 80 percent is a design target under ideal conditions, not a warranty that your pack will print 80.0 at cycle 1,000.
Two cells, one product name
GSMArena’s November retest is the cleanest paired comparison in the set, because it is the same lab, the same protocol, and two 17 Pros. The eSIM unit, model A3256, is listed at 4,252 mAh. The nano-SIM unit, model A3523, is listed at 3,998 mAh in the prose and 3,988 mAh on the charging chart. The gap from 3,998 to 4,252 is 254 mAh, 6.35 percent. GSMArena called that 6.35 percent. My division matches their percentage. Their chart label does not match their prose. I kept both.
They also wrote that Apple had said the eSIM models should deliver two additional hours of video playback versus the physical-SIM counterparts. I did not find a second set of video hours on the spec sheet I opened on 27 September 2026. That page now says iPhone 17 Pro and 17 Pro Max use eSIM and are not compatible with physical SIM cards. Trusted Reviews, writing in December 2025 about a UK unit, said that unit had a physical SIM slot. Both documents can be true if Apple changed the shipping configuration, or if the spec page describes the configuration Apple wants cited now and the reviews describe what was in the box then. I cannot settle that from a desk. What I can settle is the paired lab result.
| Measure | Nano-SIM 3,998 mAh | eSIM 4,252 mAh |
|---|---|---|
| Calls | 1394 min | 1441 min |
| Web | 759 min | 821 min |
| Video | 1367 min | 1360 min |
| Game | 614 min | 658 min |
| Active mix | 923 min | 972 min |
Component deltas, eSIM minus nano, from the minutes above:
- Calls: +47 min (3.37 percent). GSMArena printed this as 47 minutes and a 3 percent improvement. The minutes match. The percent is their rounding.
- Web: +62 min (8.17 percent). They called this the biggest gain and “over an hour.” 62 minutes is over an hour. Check passed.
- Video: -7 min (-0.51 percent). They said they repeated the test and never saw Apple’s extra two hours. The table shows a seven-minute loss, not a two-hour gain.
- Gaming: +44 min (7.17 percent). They printed 44 minutes. Check passed.
- Active Use Score: +49 min (5.31 percent). They printed 49 minutes and 5.3 percent. Check passed.
Milliamp-hours per hour of their Active Use Score are 259.9 on the nano-SIM unit and 262.5 on the eSIM unit. That is a 0.99 percent difference in draw, which is noise next to a 6.35 percent capacity gap. On this mix, the larger cell bought roughly proportional time. It did not buy a more efficient phone. Video is the exception, and it is the exception that matters, because video is the test Apple uses in the spec sheet.
If you are choosing among 17-series sizes rather than between two SIM trays, the model guide is which 17 to buy. This recomputation does not replace that. It tells you that “17 Pro” on a spec sheet is not yet a single battery. Regional inventory still matters if a physical-SIM unit is what a shop actually has, whatever the current spec page prefers to describe.
The video test that did not move
This is the load-bearing result. Capacity up 6.35 percent. Video rundown down 7 minutes. GSMArena said they repeated it. I cannot repeat it. I can refuse to sand the number off.
Apple’s streamed claim for the 17 Pro is 30 hours, 1,800 minutes. GSMArena’s eSIM video result is 1,360 minutes, 75.6 percent of that claim, a gap of 440 minutes. The local-playback claim is 33 hours. The same lab video is 68.7 percent of that claim. I am not scoring Apple as having “missed by 24 percent.” Their procedure is local or streamed playback under their conditions. GSMArena’s video leg is their video leg, and a reader comment on that news post even argued they should have used local files rather than streaming. The point of the subtraction is narrower: if a review headline treats 22 hours 40 minutes and 30 hours as the same kind of fact, the buyer is being asked to ignore a seven-hour noun collision.
| Measure | Hours as printed |
|---|---|
| Apple local | 33 h |
| Apple stream | 30 h |
| GSMArena video | 22.67 h |
| Tom's Guide web | 15.53 h |
| Tech Chap Max | 7.97 h |
The Tech Chap shootout, as printed by 9to5Mac, makes the collision larger rather than smaller. Continuous YouTube on Wi-Fi and cellular killed a 17 Pro Max in 7 hours 58 minutes. Apple’s streamed claim for that model is 35 hours. The YouTube session is 22.8 percent of the claim. Year over year inside that one YouTube protocol, the Pro Max gained 29 minutes, 6.46 percent, over a 16 Pro Max at 7 hours 29 minutes. Apple’s own Pro Max streamed claim gained 6 hours, 20.69 percent. Same direction. Not the same magnitude. A buyer who hears “six more hours” and a buyer who hears “half an hour more on YouTube” are both repeating a source. They are not repeating the same experiment.
Camera time is the practical version of this gap. A six-month camera report, including the 17 Pro camera after six months, is full of the workload that does not look like Apple’s video-playback slide: viewfinder on, screen bright, radios up, files being written. Trusted Reviews’ spot check is the only short-session video number I will put next to that, and I will not stretch it into a rundown without a label. One hour of Netflix in HDR cost 5 percent on their UK unit. A straight line through that hour empties the phone in 20 hours. That line assumes the last 5 percent costs what the first 5 percent cost. It is not a measured empty time. It sits between Tom’s Guide’s 15:32 web drain and GSMArena’s 22:40 video rundown, which is a plausible neighborhood and not a confirmation of either.
A web-surfing year-over-year, and the capacity gain it did not match
Tom’s Guide’s review page, which I re-fetched, prints 15 hours 32 minutes for the 17 Pro and 14 hours 7 minutes for the 16 Pro. That is 85 minutes, 10.04 percent. The prose calls it a marked improvement and ties the gain to the larger cell in eSIM-only configurations. The comparison table on that same page leaves battery size as N/A. So the runtime pair is a measurement I can cite. The milliamp-hour pair is an inference.
If their 17 Pro was the 4,252 mAh cell, capacity versus the 3,582 mAh 16 Pro that GSMArena and PhoneArena both print is 18.70 percent. Runtime gain divided by that capacity gain is 0.54. About half the capacity increase showed up in their web test. If their unit was the smaller cell, capacity gain versus 3,582 mAh is 11.61 percent using GSMArena’s 3,998 label, or 11.33 percent using PhoneArena’s 3,988 label, and the runtime ratio is closer to nine tenths. The review’s own sentence points at the larger cell. I am not going to pretend the N/A cell in the table is a settled 4,252.
Either reading has the same shopping implication. Apple’s spec-sheet streaming gain, 16 Pro to 17 Pro, is 36.36 percent. Tom’s Guide’s web-test gain is 10.04 percent. You can believe both labs and both claims without believing they measured one thing. The Android-to-iPhone switching guide is where this site talks about living with the platform. Battery math from a keynote is a weak reason to switch, and a weak reason to stay, until you know which protocol the number came from.
One reviewer anecdote on that Tom’s Guide page is worth keeping in its box. The writer describes a day of web, photos, YouTube, and a couple of games, with the phone near 20 percent at night, and says they did not top up. That is one person, one stretch of days, no brightness lock in the sentence. It matches the shape of “a full day for a heavy user” and it does not produce a minute count I can put in the table. I am leaving it as a qualitative note, not a sixth lab.
Charging: three labs agree at 15 minutes, then drift
Apple’s claim is specific. Up to 50 percent in 20 minutes with a 40W or higher adapter and a USB-C cable. Up to 50 percent in 30 minutes on MagSafe with a 30W or higher adapter. Footnote 8 names the adapters and says the clock starts at the Apple logo on a drained preproduction unit.
The 15-minute mark is the most reproducible number in this whole exercise. Tom’s Guide: 40 percent, Apple 45W brick. GSMArena: 40 percent, Pixel AVS adapter and, separately, a Samsung 45W adapter, same result. Trusted Reviews’ spec table: 40 percent at 15 minutes. Three labs, three charger stories, one printed percentage.
| Measure | 15 min | 30 min |
|---|---|---|
| Tom's Guide | 40 % | 72 % |
| GSMArena | 40 % | 70 % |
| Trusted Reviews | 40 % | 67 % |
The 30-minute mark is where the sheets separate. Tom’s Guide 72 percent. GSMArena 70 percent, and a full charge in 1 hour 18 minutes, with momentary peaks up to 35W and sustained power just under 30W. Trusted Reviews 67 percent at 30 minutes, and 80 minutes to full. The 30-minute spread is 5 points. The full-charge spread, on the two labs that printed one, is 2 minutes. I would not call those identical tests. I would call them close enough that “about 70 percent in half an hour, full in roughly 80 minutes, on a 40W-class brick” is a fair desk summary, and “50 percent in 20 minutes” remains a different mark.
Only Trusted Reviews published the 20-minute mark Apple actually claims. They got 51 percent with a 40W plug. Their table also says 21 minutes from 0 to 50 percent. Those two sentences agree within a minute. GSMArena said they could not get Apple’s North American 40W Dynamic Power Adapter at the time and used other AVS and PPS bricks. Their 40 percent at 15 minutes is not a miss of the 20-minute claim, because they did not publish the 20-minute point. Tom’s Guide’s review page, as I fetched it, also stops at 15 and 30.
Wireless: GSMArena said a third-party Qi2 charger hit the promised 50 percent from flat in 30 minutes, drawing up to 29W from the wall. That is one confirmation of Apple’s MagSafe timing, on one charger, not a survey of pads. If charge speed is why you are trading in, the money question is still the one in the trade-in process guide: what you get for the old phone, not whether a lab hit 70 or 72.
Trusted Reviews also lists the UK unit at 205 grams against a 16 Pro at 199 grams. Apple’s spec sheet lists the 17 Pro at 206 grams (7.27 ounces). One gram is not a battery finding. I am noting it so a reader who cross-checks weights does not think I silently picked a side.
One published diary, recomputed from the rows
Victor Hristov at PhoneArena published five days on a global 17 Pro, recorded at 100 percent battery health and 52 cycles. The table:
| Day | Drain | Screen time | Minutes | Minutes per 100% drain |
|---|---|---|---|---|
| Sunday | 100% | 5h 14m | 314 | 314 |
| Monday | 70% | 4h 16m | 256 | 366 |
| Tuesday | 75% | 4h 44m | 284 | 379 |
| Wednesday | 85% | 4h 20m | 260 | 306 |
| Thursday | 93% | 5h 30m | 330 | 355 |
Screen time across the five rows is 1444 minutes. Drain across the five rows is 423 percentage points, or 4.23 full batteries. Divide, and you get 341.4 minutes of screen time per 100 percent of drain. That is 5h 41m. The unweighted mean of the five per-day normalizations is 343.8 minutes, 5h 44m.
The article’s own summary says average screen time per 100 percent is 5 hours 20 minutes, that 80 percent of a battery is 4 hours 40 minutes, and that 10 percent of battery is 40 minutes of screen. Those three sentences do not agree with each other, and the first one does not agree with the table.
- 5 hours 20 minutes is 320 minutes. The table’s drain-weighted figure is 341 minutes. The gap is 21 minutes.
- 40 minutes per 10 percent implies 400 minutes per 100 percent, which is 6 hours 40 minutes. That matches neither 5 hours 20 minutes nor the table.
- 5 hours 20 minutes per 100 percent implies 32 minutes per 10 percent, not 40.
- 4 hours 40 minutes per 80 percent is exactly 5 hours 20 minutes scaled by 0.8. So the 80 percent line is just the 100 percent line times 0.8. It is not an independent measurement, and it inherits whatever produced the 5 hour 20 minute figure.
I am not accusing the diary of invention. The piece includes screenshots I could not read as an extra table. Those images may hold days the HTML table omits, or a weighting I was not shown. What I can say, because I added the printed rows, is that a reader who uses the summary and a reader who uses the table will plan different days. The summary is the line that will get quoted. The table is the line that can be checked. On a smaller phone, the 17e first look is a reminder that “Pro” endurance does not transfer down the lineup. This diary does not transfer cleanly even inside one article.
The capacity label on that PhoneArena page is part of the same honesty problem, at a smaller scale. They print 4,252 mAh for the US eSIM model and 3,988 mAh for the global model, and they say they used the global model. GSMArena’s prose says 3,998 for the physical-SIM cell. Ten milliamp-hours will not change your Thursday. It will change a percentage in a comparison chart if someone picks a label without saying which label. I used 3,998 when I was dividing GSMArena’s own minutes, and I used 3,988 when I was checking PhoneArena’s printed +11.3 percent. (3,988 minus 3,582) divided by 3,582 is 11.33 percent, which rounds to the +11.3 percent they printed. Their percentage matches their label. It does not match GSMArena’s prose label. Both can stay.
A ruler for cycle count, not a six-month curve
The queue wanted a degradation curve from 50 users over six months. I do not have those users. Drawing one would be fraud. What I have is Apple’s endpoint and one published early-life point.
Under a straight line from 100 percent to 80 percent across 1,000 cycles, health falls by 0.02 points per cycle. At 52 cycles, the PhoneArena unit’s count, the line reads 98.96 percent. The phone was still showing 100 percent. That is compatible with display rounding, with a pack that has not stepped down yet, and with Apple’s separate note that a battery can read under 100 percent at activation simply because of time in the warehouse. A single 100 percent reading at cycle 52 does not validate the 1,000-cycle rating, and it does not threaten it.
At 183 cycles, one cycle a day for six months, the same line reads 96.34 percent. At 365 cycles, one a day for a year, it reads 92.70 percent. I did not observe either phone. If your habit is two cycles a day, you are not on this line, and Apple already said the percentage depends on use. The hidden settings piece is the right companion if you want the charge-limit switch. This section will not pretend a slider is a study.
Service threshold remains the one Apple prints in the support flow: they talk about replacement when maximum capacity is significantly degraded, and the long-standing consumer reading of that screen is the 80 percent mark. Apple’s battery service page still tells owners that lifespan varies with use and settings, and that built-in packs should be serviced by Apple or an authorized provider. Nothing in the lab sheets I recomputed moves that advice. A phone that still shows the high 90s at half a year is what the straight line expects. A phone that is in the 80s at half a year is off that line, and the line was never a warranty.
The environmental report linked from the spec sheet says the 17 Pro battery uses 100 percent recycled cobalt and 95 percent recycled lithium. That is a materials claim, not an endurance claim. I am including it because it is on the same Apple page as the hour claims, and readers sometimes treat every battery sentence as a runtime sentence. It is not one.
What this does not tell you
What this does not tell you: It does not tell you how your 17 Pro will look at month six. It tells you how four published sheets compare, where they agree, and where a summary does not reproduce.
No handset sat on my desk. Brightness, cell signal, iOS build, case heat, and whether Optimized Battery Charging was on are uncontrolled because they were someone else’s uncontrolled variables. I cannot assign the seven-minute video regression to a SIM tray, a firmware build, or a measurement wobble. GSMArena said they repeated the video leg. Repetition reduces the chance of a one-off timer error. It does not identify a cause.
The 16 Pro hour baseline is a MacRumors transcription from launch week, re-read today, because Apple’s old product URL redirected. If Apple has edited those hours since and the edit lives only on a page I could not open, my year-over-year spec comparison inherits that gap. The 17 Pro hours came from the live spec sheet.
Tom’s Guide’s comparison article, the one that estimated 4,252 mAh next to the 15:32 result, returned 403 to the extraction tool. I cited runtime and charge percent only from the review page I fetched, where capacity is N/A. Do not treat this post as confirming that Tom’s Guide’s unit was the larger cell. The review prose points that way. The table does not prove it.
The PhoneArena screenshots may contain the weighting that produces 5 hours 20 minutes. Until those images are transcribed, the conflict stands as a conflict between the HTML table and the HTML summary, not as proof the author cannot add. I am also one diary, not a distribution. A single global unit at 52 cycles says nothing about the owner who lives on 5G in a weak market and charges to 100 percent every night.
iOS has moved since the September 2025 reviews. A drain test on launch software is not a drain test on the build you install tomorrow. Features that only show up on 15 Pro and newer is a software story, not a battery story, but it is the right caution: a year-old lab sheet is already a historical document. I did not rerun anyone’s protocol on the current build. If a later iOS build changed modem behavior or always-on display cost, these gaps move, and this post will not move with them until someone publishes a new sheet.
I did not price the phone, and I did not score the camera. Battery is one axis. The 17 Pro camera review is a different axis, and mixing them into a single “worth it” verdict is how ordinary reviews hide the protocol problem this piece is about. I also did not measure the 17 Pro Max except where a source printed it as a contrast. The Max has its own cell and its own claims. Borrowing the Pro’s 49-minute eSIM gap and pasting it onto the Max would be the same mistake as pasting 33 hours onto a YouTube afternoon.
What I would actually do with these sheets
If I were buying a 17 Pro for battery life in September 2026, I would use three numbers and ignore the rest of the adjectives.
First, the paired GSMArena result. The larger cell is real, about 6.4 percent, and it showed up in calls, web, and gaming. It did not show up in their video rundown. If your day is video, the eSIM capacity story is weaker than the keynote version of that story. If your day is mixed, their Active Use Score moved 49 minutes. That is a real gain and a modest one.
Second, the charge cluster. Forty percent in 15 minutes showed up three times. Fifty percent in 20 minutes showed up once, at 51 percent, on the lab that used a 40W plug and published that mark. I would believe the 20-minute claim for a 40W-class brick. I would not believe a comment that says every charger does it.
Third, the noun check. Thirty-three hours, 15 hours 32 minutes, and 7 hours 58 minutes can all be printed in the same week and all be “battery life.” They are not interchangeable. When a shop listing or a forum post drops the protocol, I treat the number as unlabeled and I do not use it.
I would not wait for a 50-person survey that does not exist, and I would not trust one that appears overnight with no instrument, no response rate, and no link to a form. The straight-line cycle ruler says a typical one-cycle-a-day owner is still in the mid-90s at six months if wear is even. That is a hypothesis with an Apple endpoint under it, not a panel. If your phone is already far below that ruler, the support page is more useful than this article.
The Apple Intelligence versus Gemini piece is here only as a reminder that software features have a power cost this dataset does not isolate. None of the four labs printed an on-versus-off intelligence comparison. If that is your question, this study does not answer it.
Frequently asked questions
Did you survey 50 iPhone 17 Pro owners?
No. The queue asked for that. I did not run a survey, I did not scrape a comment section and call it a sample, and I did not write fictional cycle counts. The dataset is a recomputation of Apple’s spec sheet, GSMArena’s two 17 Pro cells, Tom’s Guide’s published drain and charge marks, Trusted Reviews’ spot checks, and one five-day diary. Sample size is the number of published sheets, not 50.
Why is Apple’s 30-hour figure so far from the lab video test?
Because they are different procedures. Apple’s streamed claim is 30 hours under the conditions in its battery footnote. GSMArena’s eSIM video leg is 22 hours 40 minutes under their protocol. The gap is 440 minutes. That subtraction shows the cost of treating the nouns as identical. It is not, by itself, a failed replication of Apple’s test. The Tech Chap YouTube session on a 17 Pro Max, 7 hours 58 minutes, is a third procedure again.
Does the eSIM model last longer?
On GSMArena’s mix, yes, by 49 minutes of Active Use Score, off a cell that is 254 mAh larger if you use their prose figures of 4,252 and 3,998. Calls, web, and gaming moved with the cell. Video did not. Their video result was 7 minutes shorter on the larger cell, after they said they repeated the test. Apple’s live spec sheet, as I opened it, no longer describes a physical-SIM 17 Pro at all. Reviews from late 2025 still do. Check the unit in front of you, not the adjective in the headline.
Is 15 hours 32 minutes a better number than 33 hours?
It is a better number for Tom’s Guide’s web-surfing drain, which gained 85 minutes over their 16 Pro result. It is a worse number if you wanted Apple’s video-playback procedure. Neither number is “real battery life” in the abstract. Pick the protocol that resembles your day, and discard the other rather than averaging them. An average of 33 hours and 15.5 hours would describe no test that was run.
How fast does the 17 Pro actually charge?
Three labs printed 40 percent at 15 minutes. At 30 minutes they printed 72, 70, and 67 percent. Trusted Reviews, the only write-up I found that hit Apple’s exact 20-minute mark, got 51 percent on a 40W plug, against a claim of 50 percent. GSMArena’s full charge was 1 hour 18 minutes. Trusted Reviews’ was 80 minutes. MagSafe’s 50 percent in 30 minutes was confirmed once, on a Qi2 pad GSMArena tried, not across a charger sample.
What battery health should I expect after six months?
I do not have a six-month panel, so I will not give you a fake expected percentage from owners. Apple’s design target for iPhone 15 and later, on Support 106348, is 80 percent at 1,000 cycles under ideal conditions, and they say real percentages depend on use. A straight line through that endpoint reads 96.3 percent at 183 cycles. That line is mine. It is not a published Apple curve. One reviewed global unit was still at 100 percent at 52 cycles, which sits next to the line rather than on a measured curve.
Which number should I use when deciding to buy?
Use the paired GSMArena cells if you want the only same-lab comparison of the two batteries. Use the 15-minute charge cluster if charge speed is the question, because that is where the labs agree. Use Apple’s hours only as Apple’s hours, with the footnote attached. Do not use a forum percentage, a YouTube comment, or this article’s straight-line ruler as if it were a measurement of your phone.
What would have made this a real six-month study?
A named instrument, a logged cycle count, a charge-limit setting, a brightness note, and more than one phone, kept for the period the title claims. Fifty respondents would have required a form, a response rate, and a refusal to publish until that rate existed. I did not have any of that this week. The next honest version of the queue title is a panel someone actually runs, not a longer version of this recomputation.
Full data appendix
Every figure below is either printed by the source or produced by the conversion described in the methodology. Blank cells mean the source did not print that mark, not that I estimated it.
| Item | Value | Source | Note |
|---|---|---|---|
| Apple local video, 17 Pro | 33 h | apple.com spec sheet | Up to. Footnote 7. |
| Apple streamed video, 17 Pro | 30 h | apple.com spec sheet | Up to. Not a lab rundown. |
| Apple local video, 17 Pro Max | 39 h | apple.com spec sheet | Contrast only. |
| Apple streamed video, 17 Pro Max | 35 h | apple.com spec sheet | Contrast only. |
| 16 Pro local / streamed | 27 h / 22 h | MacRumors, 10 Sep 2024 | Live Apple 16 Pro URL redirected. |
| 16 Pro Max local / streamed | 33 h / 29 h | MacRumors, 10 Sep 2024 | Same transcription. |
| Spec local YoY, 17 Pro vs 16 Pro | +6 h (22.22%) | Recomputed | Claim versus claim. |
| Spec stream YoY, 17 Pro vs 16 Pro | +8 h (36.36%) | Recomputed | Claim versus claim. |
| Nano-SIM cell | 3,998 mAh prose; 3,988 mAh chart | GSMArena review | 10 mAh label conflict. Not averaged. |
| eSIM cell | 4,252 mAh | GSMArena, 10 Nov 2025 | Model A3256. |
| Capacity gap, prose labels | 254 mAh (6.35%) | Recomputed | Matches GSMArena’s 6.35%. |
| Nano call / web / video / game / active | 23:14 / 12:39 / 22:47 / 10:14 / 15:23 | GSMArena | Minutes: 1394, 759, 1367, 614, 923. |
| eSIM call / web / video / game / active | 24:01 / 13:41 / 22:40 / 10:58 / 16:12 | GSMArena | Minutes: 1441, 821, 1360, 658, 972. |
| eSIM minus nano, video | -7 min (-0.51%) | Recomputed | Larger cell, shorter video rundown. |
| eSIM minus nano, active | 49 min (5.31%) | Recomputed | Matches their 49 min / 5.3% summary. |
| mAh per active hour | 259.9 nano; 262.5 eSIM | Recomputed | Efficiency almost flat. |
| Tom’s Guide drain | 15:32 vs 16 Pro 14:07 | Tom’s Guide review, re-fetched | +85 min (10.04%). Cell size N/A on that page. |
| Charge, 15 min | 40% / 40% / 40% | Tom’s Guide, GSMArena, Trusted Reviews | Different bricks. |
| Charge, 30 min | 72% / 70% / 67% | Same three | 5-point spread. |
| Charge, 20 min | 51% vs Apple 50% | Trusted Reviews prose | Only lab that printed this mark. |
| Full charge | 78 min GSMArena; 80 min Trusted Reviews | Both reviews | Different cells and bricks. |
| Netflix HDR, 1 hour | 5% | Trusted Reviews table | Linear empty would be 20 h. Not a rundown. |
| Light gaming, 30 min | 7% | Trusted Reviews table | Do not merge with GSMArena’s game rundown. |
| Diary, drain-weighted screen per 100% | 341.4 min (5h 41m) | Recomputed from PhoneArena rows | Stated summary was 5h 20m. Not reproduced. |
| Straight line at 183 cycles | 96.34% | Recomputed from Apple endpoint | Not an observed cohort. |
| Tech Chap YouTube, 17 Pro Max | 7:58 vs 16 Pro Max 7:29 | 9to5Mac citing The Tech Chap | +29 min. Different model from the 17 Pro rows. |
Primary pages used: Apple 17 Pro specs, Apple 17 Pro product page, Support 106348, Support 101575, battery service, September 2025 environmental report, GSMArena review, GSMArena eSIM retest, GSMArena battery-test method, Tom’s Guide review, Tom’s Guide charging piece, Trusted Reviews, PhoneArena diary, 9to5Mac, The Tech Chap video cited by 9to5Mac, MacRumors 16-series hour table, MacRumors 17 Pro roundup.
Conclusion
I did not find a 50-person battery panel, and I did not write one. What the published sheets support is narrower and, I think, more useful. The 17 Pro’s larger eSIM cell is real. On the only paired protocol, it bought 49 minutes of mixed use and bought nothing in the video leg, which is the leg Apple uses to advertise the phone. Three labs agree that 15 minutes on a fast brick is about 40 percent. One lab reproduced the 20-minute, 50 percent claim. A five-day diary’s summary does not add up from its own table. A straight line through Apple’s 1,000-cycle endpoint is a ruler, not a cohort.
If you came for a personal six-month curve, this is the wrong piece, on purpose. The settings that change a day you are already living are in the battery settings guide. The cycle screen, and what Apple will and will not replace, is in the battery-health guide. The data in the appendix is the part I will stand behind, because every row can be checked against a page that was open on 27 September 2026.