Initial benchmarks show iPhone 13's A15 chip offers a roughly 15% improvement for GPU performance and 10%-18% improvement for CPU performance over A14
Geekbench 5 Score 14216 Metal Score Geekbench 5.4.1 for iOS AArch64 Result … Dashia Starr / CNET : AT&T's ‘free’ iPhone 13 trade-in offer is now available: What you need to know Oliver Haslam / iMore : The majority of new iPhone 13 Pro, Pro Max orders are delivering next month Aleksandar Anastasov / phonearena.com : The iPad mini 6 has the iPhone 13's chip, but there's a catch James Peckham / TechRadar : New iPad mini 2021 doesn't support the same 5G tech as iPhone 13 or iPad Pro MK Tomar / Micky : The A15 in iPhone 13 has 21 percent more power than A14 Nica Osorio / International Business Times : New Test Scores Unveil iPhone 13 Pro Unmatched CPU, GPU Performance Darragh Murphy / Laptop Mag : iPhone 13's A15 Bionic shows off upgraded benchmarks — Pixel 6 Pro falls behind Sanuj Bhatia / Pocketnow : A15 Bionic chip inside the iPad mini 6 clocks slower than the one inside iPhone 13 Killian Bell / Cult of Mac : iPad mini's new A15 Bionic chip isn't as fast as iPhone 13's Benjamin Mayo / 9to5Mac : Apple Store is down ahead of iPhone 13 preorders Tweets: @universeice : Apple A15 Geekbench score, only one word left to Android: despair. https://twitter.com/... Maynard Handley / @handleym99 : @reneritchie ... I do agree with the point that, *for iPhone*, transistors spent elsewhere on the SoC than the big cores are probably a better fit to the product. Which is why I hypothesize that what has REALLY changed is: new cores debuting in the iPhone. Neil Cybart / @neilcybart : As I continue to work through my Apple event notes and more information about the features become available, it's becoming clear that the early hoopla around iPhone and Apple Watch chips was partially driven by FUD. Bruce Burke / @bruceqburke : @reneritchie This is the real iPhone Pro Max this year. 🚀 Who wants a folding Samsung tablet for $1000, when you can have an iPad mini for $500? 🙌 It's not too late for back to school and will be huge for the holidays. They will sell a ton of these this year. Ben Bajarin / @benbajarin : @dylan522p @reneritchie I think it is worth noting the 52% increase in GPU being the largest since the jump from A8-A9. So clearly they put that transistor budget to something else they felt was more important to the core experience. Average GPU perf yoy increase was 30% last four chips. Dylan Patel / @dylan522p : @reneritchie @BenBajarin My comments in the article are solely on CPU performance and architecture. This is the first generation without IPC increases. That has been proven true. I praised other aspects of the SOC. Maynard Handley / @handleym99 : @dylan522p ... What's clear is that there is a CHANGE. (1) Calling it a stumble (as opposed to, eg, my hypothesis of a pivot to - 2yr CPU cycle - new cores *debuting* on the Mac side) and (2) then proffering an explanation (engineers leaving) is where you are receiving pushback... Alex Russell / @slightlylate : Imagine if we were allowed to ship a quality browser on that sweet, sweet Apple silicon... 🤔🤔🤔 Maynard Handley / @handleym99 : @reneritchie ... That's not correct Renee. Even without process improvement, there is a constant stream of µArch improvements resulting in higher IPC and/or lower energy, and there is no reason to believe those have hit a wall. (Hell I can list 10 things they could be doing in new designs.) Nati Shochat / @natisho : “These scores represent a roughly 10% increase in single-core performance and 18% increase in multi-core performance compared to the A14 Bionic in the iPhone 12 lineup.” The CPU improvements YoY are starting to be in the mid teens. https://www.macrumors.com/... Alex Russell / @slightlylate : If the benchmark previews are right, the Pixel 6 will get TROUNCED both by the iPhone 13 as well as last year's A12. Embarrassing. Again. https://twitter.com/... Apple Tomorrow / @apple_tomorrow : I remember seeing a couple of guys saying “oMg aPpLe cAnT mAkE a bEtTeR cPu” https://twitter.com/... Rene Ritchie / @reneritchie : @BenBajarin @dylan522p Yeah, if you look back at everything Apple's said for years, perf is never a goal but a result of efficiency, and absent process shrink this year, you have standard generational improvements but significant transistor budget going into efficiency and non-core features. Alex Russell / @slightlylate : Somewhat less confused now: https://www.macrumors.com/... Apple continues to squeeze performance by doing what Qualcomm et. al. will not: building caches big enough to keep the beast fed. That costs mm^2, but the results speak for themselves: a consistent multi-year lead. https://twitter.com/...
Context & Ripple Effects
Apple’s A-series story had already been one of sustained single-thread gains: analysis of A14 found performance had nearly tripled over five years. The initial iPhone 13 results put the A15’s year-over-year advance into that continuing cadence.
The early scores are only one view of the chip. A subsequent A15 workload review found gains varied widely by task while highlighting efficiency, making the benchmarks a useful launch signal rather than a complete performance verdict.
First-order effects
- Apple can market iPhone 13 on a measurable CPU and GPU lead over A14-based iPhones, with the reported CPU uplift varying by benchmark result.
- iPhone 13 buyers receive a higher performance baseline, while the different CPU results make workload-specific testing more important for comparing A15 devices with their predecessors.
Second-order effects
- Apple’s performance narrative shifts from a single headline score toward efficiency and task-level behavior, as the later A15 review documented materially different gains across workloads.
- The A15’s reported GPU result joins analysis of its unusually large GPU advance, increasing the importance of graphics performance alongside CPU comparisons in annual iPhone-chip evaluations.
Third-order effects
- Apple’s mobile-silicon cycle is increasingly judged by the balance of CPU, GPU, and efficiency gains rather than by one benchmark; later coverage of A16’s marginal benchmark lead over A15 underscores that annual progress need not be uniform across generations.
The trend: Apple’s A-series progression is moving toward more differentiated gains by workload and component, with efficiency becoming a central measure alongside peak benchmark performance.