Apple reporting record
Across 133 months, 728 leaks about unreleased Apple devices, their features and their development status trace to a named source. 179 of them belong to one person.
Who reported Apple news, and how the record was checked. Read every report, newest first
216 of 257 checked details were right or partly right
Not part of that number
Finding 01
Mark Gurman is credited with 179 of the 852 leaks in this edition — more than two times the next source. No other beat we track is this concentrated, and it is the single fact that most changes how an Apple rumor should be read.
Finding 02
42 sources have broken two or more stories here; 19 have broken exactly one. A single correct leak is not a track record, which is why the chart plots the repeat sources and the table still lists everybody.
Finding 03
Where the outlet we read was relaying somebody else, we record a link to the ORIGINAL report as well: 271 rows carry one, and on 46 of those we have checked the quoted sentence against the page itself. 797 of the 852 reached us through an outlet other than the one that broke them.
Every figure on this page is computed from the edition being rendered. If the table below shows a different number from a sentence above it, that is a bug and not a rounding choice.
A few examples of how pre-launch reporting compared with what the company later announced.
Apple iPhone Fold
Gurman names Apple's foldable the iPhone Duo and puts it at roughly $2,000, hours before the launch event.
Right
Apple iPhone Fold
Instant Digital says the foldable iPhone reaches China later than the rest of the world, because it is eSIM only.
Wrong
Apple iPhone Fold
Gurman says the memory shortage moved Apple's foldable target from $1,999 to $2,199, with large storage near $3,000.
Wrong
iPhone 18 Pro
pdfu found iOS 27 code that varies lens correction with aperture, which only pays off on a lens that can change one.
Right
Every leak, with its date, the source it traces to and the article the attribution was found in.
Chart and table show the live index, 852 leaks.
Bar length is the number of leaks. Every named source is shown. Select any bar to read the leaks behind it.
852 leaks live · last frozen edition v9, 82 leaks (2026-07-31) — download CSV
| Source | Type | Leaks | Share |
|---|---|---|---|
| Mark Gurman | Named reporter | 179 | 25% |
| Ming-Chi Kuo | Leaker or analyst | 84 | 12% |
| Bloomberg | Publication | 53 | 7% |
| Fixed Focus Digital | Leaker or analyst | 52 | 7% |
| Jeff Pu | Leaker or analyst | 39 | 5% |
| Digital Chat Station | Leaker or analyst | 32 | 4% |
| Instant Digital | Leaker or analyst | 27 | 4% |
| The Information | Publication | 23 | 3% |
| Sonny Dickson | Leaker or analyst | 19 | 3% |
| MacRumors | Publication | 19 | 3% |
| 9to5Mac | Publication | 14 | 2% |
| TrendForce | company | 14 | 2% |
| danbo | Leaker or analyst | 13 | 2% |
| Macworld | Publication | 12 | 2% |
| Ross Young | Leaker or analyst | 12 | 2% |
| Kosutami | Leaker or analyst | 11 | 2% |
| Ice Universe | Leaker or analyst | 11 | 2% |
| DigiTimes | Publication | 10 | 1% |
| Majin Bu | Leaker or analyst | 9 | 1% |
| The Elec | Publication | 9 | 1% |
| Tim Long | Leaker or analyst | 7 | 1% |
| Lakhwinder Singh | Leaker or analyst | 5 | 1% |
| @earlyappleleaks | Leaker or analyst | 5 | 1% |
| WayLabs | Leaker or analyst | 4 | 1% |
| pdfu | Leaker or analyst | 4 | 1% |
| ETNews | Publication | 3 | 0% |
| ShrimpApplePro | Leaker or analyst | 3 | 0% |
| Sam Henri Gold | Leaker or analyst | 3 | 0% |
| Jon Prosser | Leaker or analyst | 3 | 0% |
| World Leaks | Leaker or analyst | 3 | 0% |
| ldt (@madeby_ldt) | Leaker or analyst | 3 | 0% |
| Economic Daily News | Publication | 3 | 0% |
| The Mac Observer | Publication | 3 | 0% |
| Dan Ives | Leaker or analyst | 2 | 0% |
| Commercial Times | Publication | 2 | 0% |
| Digital Trends | Publication | 2 | 0% |
| KB Securities | company | 2 | 0% |
| Kim Young-ho | Named reporter | 2 | 0% |
| AppleInsider | Publication | 2 | 0% |
| Jefferies | company | 2 | 0% |
| DealSite | Publication | 2 | 0% |
| United Daily News | Publication | 2 | 0% |
| NFCW | Publication | 1 | 0% |
| Asher Dipprey | Leaker or analyst | 1 | 0% |
| Cult of Mac | Publication | 1 | 0% |
| Dan Nystedt | Leaker or analyst | 1 | 0% |
| Gizmodo | Publication | 1 | 0% |
| @LusiRoy8 | Leaker or analyst | 1 | 0% |
| Lanzuk (yeux1122) | Leaker or analyst | 1 | 0% |
| UploadVR | Publication | 1 | 0% |
| Charlene Chen | Named reporter | 1 | 0% |
| Ninelu Tu | Named reporter | 1 | 0% |
| Tim Culpan | Named reporter | 1 | 0% |
| Edison Lee | Leaker or analyst | 1 | 0% |
| Macwelt | Publication | 1 | 0% |
| Mactracker | Leaker or analyst | 1 | 0% |
| Fried White Rice (Weibo) | Leaker or analyst | 1 | 0% |
| Vodafone | Leaker or analyst | 1 | 0% |
| Nikkei Asia | Publication | 1 | 0% |
| never_released | Leaker or analyst | 1 | 0% |
| yeux1122 | Leaker or analyst | 1 | 0% |
This is not a ranking. Open a profile to see what we researched and, where enough outcomes exist, how it held up.
Every leak counted here revealed something Apple had not announced about a product: a device, feature, design, timing or project status. Mark Gurman of Bloomberg broke 179 of the 852, more than two times the next source.
Strip out Gurman and what remains is not a second tier so much as several different trades.
Supply-chain reporting. Ming-Chi Kuo, an analyst, and DigiTimes, a publication, both cover component choices, orders and production timing here.
Independent leakers. Instant Digital and Kosutami post to social platforms and are picked up by publications afterwards. Ten leaks here originate with them.
The Korean trade press. The Elec reports displays and panels for a Korean readership. It accounts for nine leaks here, both concerning display technology.
The publications reading code. MacRumors, 9to5Mac and Macworld account for forty-five leaks between them, and most came from examining Apple software, backend material and public regulatory filings. It is a different craft from Gurman’s and a more checkable one. When Aaron Perris found an Apple Watch image in an iOS 26 beta carrying a resolution matching no shipping model, the evidence sat in a file anyone could open.
All 852 leaks behind the chart, in order. Filter by source, or click a circle above to see one source on its own.
How to read this: We count only details we can meaningfully compare with what Apple later announced or shipped. Still-open items stay separate. Reporter history is context, not the probability that the next report is true.
What “traced” means here. For each leak we confirmed who the article carrying it credited with originating it. That is a check on attribution, not on accuracy: it does not mean the underlying claim was later confirmed by Apple, and several of these have not been.
On who was first. Credit goes to the earliest originator identified in the cited coverage that we could confirm. This is not an exhaustive search of every publication, newsletter, social post, language and deleted source worldwide, and we do not claim it identifies the first person anywhere to publish a given fact.
What one leak means. One leak is one substantive revelation about one product by one source. Not one fact: a report that a headset has a new chip and a new strap is one leak, not two. Not one article either: where several publications carried the same revelation, or one newsletter covered three separate products, we count the revelations rather than the write-ups. That keeps the count independent of how any outlet packaged someone else’s work.
The same rule for every format. A source earns one leak for each product about which it reveals something genuinely new, whether that arrives as a newsletter, an analyst note, a social post or an obtained file. A document is not split fact by fact, and codenames, specifications and timing about the same product are one leak, not three. A later article counts only where it adds a new revelation rather than re-explaining the same material.
Products, features and project status. A substantive revelation about a product, feature or development decision Apple has not announced counts, even where the hardware already ships. That includes unreleased software features, product timing, projects in development, and sourced reporting that a project was shelved or cancelled. Software unrelated to Apple hardware does not count.
Forecasts are not leaks. Sourced negative reporting counts: a project shelved, a successor not in development. A reporter saying they personally do not expect something does not.
Product placed by the company is not a leak. Unreleased hardware worn in public by athletes and other partners ahead of launch is marketing, and earns no credit here. A journalist finding the same product in a regulatory filing is reporting, and does.
Why the link usually points elsewhere. The source credited with a leak is rarely the publication we link to. The link goes to the article in which we found the attribution, which is usually a second publication reporting on the first: 797 of the 852 leaks here. The other fifty-five are marked own article: the publication credited broke the story itself. Where we could locate the credited publication's own article, a second link is given. 271 leaks carry one, and 46 of those sit behind a subscription and are marked.
What does not. Apple's own announcements, keynote and press-release material, reviews, hands-on pieces, deals, buying guides, and roundups restating what was already known.
Patents are excluded. A filing describes something a company has considered, not something it will ship. It is not counted here until somebody finds it and reports it. Apple patents published in this period are not counted here.
What this ranks. Sources: people and publications alike. Where citing coverage consistently names an individual reporter, the reporter is credited; otherwise the publication is. The chart draws every named source, including the nineteen holding a single leak each: on a page about who gets credited, a chart with a minimum would hide exactly the rows worth checking.
Where the counts are conservative. Some of these publications sit behind subscriptions, so reports not cited by accessible coverage may be missing. Leaks that first appear on X or Weibo are counted only where a publication picked them up. The source counts are therefore conservative; percentages and relative leads describe only the 728 leaks in this dataset that trace to a named source.
How this was checked. The dataset was audited row by row before publication and rebuilt as duplicate, scope and attribution rules were applied consistently.
Corrections. Every bar opens to the leaks behind it, with dates and links, so any figure on this page can be checked. If something here is wrong, or a leak is credited to the wrong person, write to hey@idevice.com and we will fix it and say that we did.
Corrected 3 August 2026. The fourth statistic at the top of this page counted the runner-up twice. It read “12 other named sources share the remaining 43” — but the figure beside it already named MacRumors separately, so MacRumors’ ten sat inside both, and the four numbers did not add up to the total. The row now excludes both sources named to its left and leads with the leak count, so the figures partition the total exactly. No leak, credit or date changed; the error was in the summary line, not the data. The downloadable edition is unaffected.
Leaks since October 2015, the most recent from October 2026. Last updated 27 October 2026. The index updates continuously as leaks are added and corrected; downloadable editions are published periodically.
