As we write this, Apple's "big week" is just hours away. No keynote, no stage, but a product lineup that promises to shake things up: an entry-level MacBook with an iPhone chip, the long-awaited iPhone 17e, new iPads, and a refreshed MacBook Air. Meanwhile, Siri is still hiding backstage. There's a lot to cover, so let's get into it.
The Week Apple Dresses Up in Color
March 2nd kicks off what Apple has planned as its announcements week, with press releases rolling out through March 4th, when in-person events in New York, Shanghai, and London wrap things up. The menu is substantial:
- Entry-level MacBook with A18 Pro chip: the undisputed star of the week. Available in colors — yellow, blue, pink, silver — with a 12.9–13-inch display and an estimated price of $599 to $799.
- iPhone 17e: A19 chip, MagSafe finally included, a 48MP rear camera, and a price that stays at $599.
- iPad Air with M4 and standard iPad with A18 — the latter supporting Apple Intelligence for the first time in the base lineup.
- MacBook Air with M5, with multi-core performance improvements of up to 20% over the M4.
- MacBook Pro with M5 Pro and M5 Max for those who need everything turned up to eleven.
A big week. And for us, the developers, the most relevant piece is undoubtedly the first one on that list.
The Mac That Could Change Who Develops for Apple
Let's be blunt: one of the biggest barriers to getting started with Apple development has never been the learning curve — it's been the price of the entry machine. For years, the minimum recommended option has been the MacBook Air — a fantastic laptop, but one that starts above $1,100 in most markets.
That's about to change. The new MacBook with the A18 Pro brings the same chip as the iPhone 16 Pro: 6-core CPU, 6-core GPU, second-generation 3nm process, and 8GB of RAM. In benchmarks, the A18 Pro matches the M3 Ultra in single-core performance and outperforms the M1 in multi-core — yes, the same chip that launched the Apple Silicon era in the Mac — in a laptop that costs under $800.
Is it a machine for demanding professional users? No, and we should be honest about that. No Thunderbolt — standard USB-C only — a smaller display, and no ProMotion. But for running Xcode, compiling real projects, working with simulators and Previews, debugging, and publishing to the App Store, it's more than enough. Apple Intelligence works. The full development toolchain works.
Alongside the Mac mini — still the best desktop option in the Apple ecosystem for the price — we now have two genuine entry points into native development at prices that, for the first time in a long while, can seriously compete with alternatives on other platforms. If you know someone who has been wanting to start Apple development but found the initial investment too steep, this week might be their moment.
iOS 26.4: The Update That Arrived Without Its Star
The first iOS 26.4 beta landed in developers' hands in mid-February, and the community went through it with a fine-tooth comb hoping to find signs of the new Siri. The verdict: nothing. The Siri that was supposed to understand personal context, search through old messages, and control third-party apps by voice is nowhere to be found.
According to Bloomberg, internal testing revealed serious issues: Siri responds too slowly on complex requests, and certain types of queries fail to process correctly. The team had to pump the brakes. The new Siri might arrive in iOS 26.5 — spring — or could be held back until iOS 27 in the fall. It wouldn't be the first time Siri has slipped, but the timing stings more than usual when competitors have been demonstrating exactly what Apple promised for months now.
On our podcast, we spent 36 minutes dissecting what happened and what it means for the future of Apple Intelligence. You can listen on YouTube or on Cuonda via Apple Coding Daily.
That said, Siri going missing doesn't mean iOS 26.4 arrives empty-handed. The update does bring something meaningful for developers: a new on-device Foundation Model — a substantially improved local LLM — replacing its predecessor. This model is the backbone of everything Apple is building toward: the engine that will process personal user context, execute commands in apps through App Intents, and serve as the foundation for the visual UI interpretation we saw in the Ferret-UI research papers. The architecture is coming together, piece by piece.
This has a concrete implication if you already have apps using the Foundation Models Framework: Apple has asked developers to review their prompts and implementations against the new model to make sure behavior hasn't shifted in unexpected ways. And the reason behind that request is fundamental, not bureaucratic: AI is non-deterministic. Unlike a conventional code function, the same input doesn't always produce the same output. An LLM calculates the most probable next token in a vector space at each step, and that probability distribution can shift dramatically between model versions depending on their training base and data. A prompt that worked perfectly with the previous model may behave differently — sometimes better, sometimes unexpectedly — with the new one. Check before it hits production.
AI Will Always Need Someone at the Wheel
This context of delayed promises makes an article from Julio César Fernández even more worth reading. Using empirical data — not opinions, not predictions, but concrete data on the current technical limitations of AI systems — he makes the case for why artificial intelligence will continue to need skilled humans behind it. And why the more you know, the more valuable you'll be.
It's available on X, on LinkedIn, and on Be Native. Especially recommended if you've been wondering whether it's worth continuing to invest in technical knowledge while AI keeps advancing.
And speaking of what AI can already do: Apple had a very active February on the research front. On February 17th, they published two papers at machinelearning.apple.com worth keeping an eye on.
Ferret-UI 2 is a multimodal model that understands user interfaces across multiple platforms — iPhone, Android, iPad, web, and Apple TV — combining high-resolution perception through adaptive scaling with GPT-4o-generated training data. But the real surprise of the month was Ferret-UI Lite: a compact 3 billion parameter agent designed to run entirely on-device, with no cloud, no network latency, and no additional cost. It can locate UI elements with 91.6% accuracy on ScreenSpot-V2 and autonomously navigate interfaces. The connection to the Siri work is obvious — and promising. Full details at machinelearning.apple.com.
Swift in the Browser: BridgeJS Reaches MVP
The Swift for WebAssembly workgroup published its February summary, and there's good news on several fronts.
The biggest: BridgeJS has reached MVP status in JavaScriptKit 0.44.1. This new interoperability layer between Swift and JavaScript lets you declare APIs using Swift macros and automatically generates bidirectional bridge code. Mark your Swift code with @JS to make it callable from JavaScript; use @JSClass, @JSFunction, @JSGetter, and @JSSetter to bring JavaScript into Swift. You can even generate bindings directly from TypeScript files. Type-safe, efficient, and finally ergonomic.
In parallel, WasmKit is progressing on Component Model support — WebAssembly's modularity standard — and ElementaryUI now includes @FocusState, a Group view, and a FilterModifier with blur, saturation, and brightness, along with general performance improvements. Full details on the Swift Forums.
Server-Side Swift: Keeping the Momentum
On February 16th, Jay Herron and the GraphQLSwift organization announced two new packages on the Swift Forums: graphql-vapor and graphql-hummingbird. The goal: expose a full GraphQL API — with WebSocket subscriptions, full GraphQL-over-HTTP spec compliance, and an integrated GraphiQL IDE — with a single call in your router:
routeBuilder.graphql(schema: schema) {
// createContext()
}The framework handles the specialized encodings, error-to-HTTP-status mapping, and subscription protocols. Nothing else to worry about.
Meanwhile, the Vapor team published their roadmap for Vapor 5: full structured concurrency (goodbye forever, EventLoopFuture), integration with Hummingbird's HTTP server, first-class support for OpenAPI, gRPC and async streaming, and complete observability from day one. Vapor 5 commits to 3 years of active support. Server-side Swift is moving in the right direction.
For Developers: SwiftUI, Liquid Glass, and Apple's February Content
February was generous in official developer content. A few things worth catching up on before the update season picks up:
- The foundational SwiftUI sessions from February 10th are still available on video, for beginners and experienced developers alike.
- The "Let's talk Liquid Glass" sessions continue into March. If you haven't explored
glassEffect,GlassEffectContainer, andglassEffectIDyet, now's the time. - New documentation on Foundation Models Framework performance optimization covers how to analyze token consumption and reduce response times in apps using Apple's on-device 3B-parameter model.
Everything available at developer.apple.com.
Marc Hervera and the Swift Developer Program 2026
Some news we're particularly excited to share: we've opened enrollment for the Swift Developer Program 2026, the definitive training for becoming a professional Apple developer, taught this edition by Marc Hervera — one of the most experienced and respected professionals in the Apple development world.
If you don't know Marc yet, now's the perfect excuse: catch the full interview on YouTube, where Julio César talks with him about his career from the Objective-C days to the AI era. If you prefer audio, it's also available on Cuonda via the Apple Coding podcast. A conversation well worth a relaxed listen.
Full program details at acoding.academy.
Gabhel Studios: The Team Behind Be Native
Finally, we'd like to formally introduce Gabhel Studios (gabhel.com), the development studio behind Be Native and many other apps in the Apple Coding ecosystem, with over 15 years of Apple development experience.
Gabhel Studios offers native iOS app development with Swift and SwiftUI — always native, never cross-platform frameworks; AI consulting, analyzing real workflows and identifying AI tools with actual, measurable impact; and legacy app modernization, including Swift 6 migrations with strict concurrency, Liquid Glass adaptation, and European Accessibility Directive compliance.
Their philosophy fits in four words: "Native code or nothing." More than 30 years in real-world projects, over 10 years specialized in AI consulting, and a refreshing honesty that we deeply appreciate: if they can't solve it, they don't charge. We're proud to work alongside a studio like this, and if you have a project in mind, we encourage you to check them out.
That's All for Now
The week ahead is loaded. New Macs, new iPhones, new iPads — and with them, new entry points to an ecosystem that remains, in our view, the best place for a developer to work. Swift isn't slowing down: it's on the server, it's in the browser, it's on the device. And we'll be here covering all of it.
If you have any comments or just want to share something, you know where to find us.
See you soon, and until then: greetings and good Apple Coding.
Thanks for reading our Newsletter! See you on Be Native.