Apple A18 and A18 Pro: The Complete Guide

Both share the same CPU naming — “6-core CPU with 2 performance and 4 efficiency cores” 34 — but differ in GPU core count, one full core apart, a distinction Apple states directly rather than leaves implied.

Apple’s iPhone 16 Newsroom release describes the 16-core Neural Engine with exactly that phrase: “optimized for large generative models” 1 — a framing Apple had not used for any earlier A-series Neural Engine description found in this site’s research.

Apple’s own performance claims for A18, versus A16 Bionic: CPU “30 percent faster... and faster than all the competition,” running “the same workload with 30 percent less power than A16 Bionic”; GPU “up to 40 percent faster and 35 percent more efficient than A16 Bionic”; Neural Engine “runs ML models up to 2x faster than the A16 Bionic chip” 1. For A18 Pro, versus A17 Pro: CPU can “run the same workload as the previous generation 15 percent faster while using 20 percent less power”; GPU “up to 20 percent faster”; ray tracing “up to 2x faster”; and “a 17 percent increase in total system memory bandwidth — the highest ever in iPhone” 2. Neither chip’s transistor count nor Neural Engine TOPS figure is disclosed by Apple in the material found.

Figures per Apple’s iPhone 16 specs page 3 and iPhone 16 Pro tech-specs page 4, corroborated by both Newsroom launch releases.

A18 Pro’s release adds: “featuring a new architecture with smaller, faster transistors” 2 — Apple’s own description of what “second-generation” means for this node, without a specific transistor count to go with it.

This is the first generation where this site’s research confirms zero shared silicon between the base and Pro iPhone chip within the same year — a genuinely different die design for each, not a binning split of one design.

No official Apple source lists a standalone chip price for A18 or A18 Pro.

Note the different comparison baselines Apple used: A18’s figures are measured against A16 Bionic (two generations back, since A18 has no direct predecessor of the same tier-naming), while A18 Pro’s figures are measured against its direct predecessor, A17 Pro. Both are company-published comparisons.

A18/A18 Pro established the fully-split naming convention (distinct base and Pro chip names) that A19/A19 Pro and A20/A20 Pro (where announced) would continue.

It would be easy to assume Apple reserves its most AI-forward marketing language for the Pro chip. Instead, the “optimized for large generative models” phrase 1 appears in the base A18’s own Newsroom release — the same generative-AI framing applied evenly across both the cheaper and more expensive iPhone that year.