M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agents

By Admin Test

The M5 Ultra Mac Studio.

For the past few days, I’ve been testing the (currently) top-of-the-line M5 Ultra Mac Studio with 256 GB of RAM.

I’ll cut to the chase: the M5 Ultra Mac Studio is a dream machine for local AI agents. This computer makes it possible to run personal assistants powered by local models with great performance and no additional cloud costs. If you’ve been skeptical of testing OpenClaw or Hermes Agent with local models because they’d never be even remotely near the intelligence and speed of cloud ones, this Mac will change your mind about that.

Since last Thursday, I’ve been comparing this Mac Studio to its predecessor, the M3 Ultra with 512 GB of RAM, as well as my own desktop gaming PC with an RTX 5090 inside. For its size, price, thermal performance – not to mention Apple’s approach to unified memory – the M5 Ultra Mac Studio has fundamentally changed how I think about models running locally and what they can enable now. A 5090, of course, still has an edge over the M5 Ultra thanks to its higher memory bandwidth. But considering the sheer size of my PC build, as well as its heat and noise, I would prefer an M5 Ultra Mac Studio any day. It also happens to be a Mac, with an operating system that looks nice and doesn’t suck, plus a vibrant app ecosystem. (Windows fans, I’m sorry, but Microsoft software will never get my sympathy.)

As I’ll explore in this article, running the latest Qwen3.8-Flash-Next model on the M5 Ultra Mac Studio has been so nice and fast, I’ve made it my default in both Open Minis for iOS and Hermes Agent. That’s right: the personal assistants I use the most – more than Siri AI, in fact – are now entirely powered by a model running locally on a Mac Studio. Furthermore, thanks to the M5 Ultra’s faster GPU and higher memory bandwidth, these agents start responding more quickly, stay fast at larger context windows, and can run long, multi-turn loops without slowing to a crawl as the session grows. Because of this, I’ve also been using local models in the Codex app on my Mac – either as main threads or subagents orchestrated by GPT-6 Astra – and I’ve had a great experience doing so.

Local subagents running in Codex on the M5 Ultra Mac Studio.

I should note upfront that I’m not an AI developer by trade: I do not train or fine-tune models. I’m a tinkerer at heart, and I’ve been playing around with local AI models for over a year at this point. This summer, I went all-in on local AI usage for a big project I was working on, which I will explain in the following section.

My goal with this article is to provide you with a mix of two things: numbers and visualizations based on the (many) tests I’ve run over the course of four days, and an explanation of my practical use cases for local AI applied to my workflow and how I get things done for MacStories.

Let’s dive in.

TestFlight Fixes and Wishes

By John Voorhees

Just as I was heading into the final stretch of review season, TestFlight was updated to reverse an update I wrote about in July. That update added search, which was one of my longstanding wishes for TestFlight. However, for some users, the update also changed the sorting of betas from reverse chronological order to alphabetical, including active betas alongside expired ones. As someone who reviews a lot of apps, that made TestFlight barely usable.

With the latest update, I’m happy to report that the sorting order has been restored to reverse chronological order. The most recently updated betas appear at the top of the list, and when you want to find something older, search is a great option. Returning to reverse chronological order may seem like a small thing, but it’s made a world of difference in keeping up with the apps I test.

Search was near the top of my list of TestFlight wishes for years, and its addition made a meaningful difference to the app’s usability. But I’ve had a lot of TestFlight wishes over the years that are scattered across multiple articles, so I thought I’d share what else I’d like Apple to tackle with TestFlight:

Performance. I test a lot of apps, which slows TestFlight down on every Apple platform. The issue isn’t as severe today as it was when I wrote about it in 2023, but the slowness is still bad enough to make the app hard to use. Whatever the issue is, I’d love to see the list loading time reduced and scrolling through a list of apps become as smooth as it is in the App Store.
Add filtering, sorting, and folders. Search is a step in the right direction, but allow me to sort by date, developer, category, and other metadata. I’d also like to be able to filter by installation status, so I can just see the betas I’ve installed or haven’t installed yet. User-defined folders for organization would be a nice bonus too.
Hide inactive and expired betas. I’ve thought about TestFlight’s expired betas section a lot and can’t think of a single reason it exists from a tester’s perspective. Let me hide or exclude expired and withdrawn betas.
Fix TestFlight emails and automatic updates. TestFlight email notifications can only be turned on or off on a per-beta basis. Adding a global switch would help a lot with managing these messages. Also, automatic updates of betas have never worked on my iPhone, and I’ve talked to others who have the same problem.
Restore usable betas after a device migration. Apple’s migration from one device to another has gotten much better over the years, but it still does not migrate TestFlight betas, leaving me with a bunch of broken apps on my iPhone every fall.
Indicate when a beta has been released on the App Store. This strikes me as easy to request, though I’m sure is difficult to implement, but I’d love to know when a beta has been superseded by an App Store version, so I can grab the stable release if I want and know I can publish a review about an app without having to check the App Store repeatedly myself.

The bottom line is that while TestFlight is once again usable, it could be much better. I recognize that TestFlight isn’t an app used by most people, but like Apple’s own OS beta testing program, TestFlight betas are an important part of developers’ workflow. They provide QA feedback and a promotional channel that the friction encountered by TestFlight users undermines.

I’m glad TestFlight continues to get meaningful updates. I just hope it gets more attention going forward.

watchOS 27: The MacStories Review

By John Voorhees

I haven’t been this excited about an Apple Watch in a long time. Last week, Apple introduced the Apple Watch Series 12 and Apple Watch Ultra 4, which will be available at the end of the week. The new Watches feature the S11 chip, which enables a bunch of new features, some of which won’t debut until later this year. At the same time, Apple is dropping support for more Watches than usual.

What we’re witnessing is a reboot of the Apple Watch and watchOS forced by advances in AI. The Apple Watch and its OS found their niche years ago, freeing us from pulling out our iPhones for notifications, tracking our health and fitness, and enabling the occasional Siri request. It was a good and comfortable spot for what by all accounts has been one of the most successful wearables to date, but watchOS also felt stuck in a rut.

Every year we got a few new workout types, a new watch face, and a few other features, but the OS didn’t advance much. What’s different this year is that watchOS 27 is breaking with slower hardware and looking forward to the substantially more powerful S11 chip, which will allow the Apple Watch to expand into all-new areas.

What this transition also means is that there is not just one set of features to track in watchOS 27 this year but three:

those that arrive today for anyone with a compatible Apple Watch,
those that arrive at the end of the week that are exclusive to the new Apple Watch Series 12 and Ultra 4, and
those that are coming later this year.

That makes this more than just a review. It’s also a preview of things to come and a roadmap for where the Apple Watch is heading.

There’s more than usual to cover this year, so let’s dive in and check it out.

macOS 27: The MacStories Review

By John Voorhees

Every OS update has a story to tell. With macOS 27 Golden Gate, that story captures a moment when Apple is faced with multiple challenges, many of which are tied to the rise of large language models.

One of those challenges is obvious. Apple was caught off guard by chatbots, overpromised a response at WWDC in 2024, and has been playing catch-up ever since.

But this isn’t simply a story of missteps or wasted opportunities. There’s more to it than that.

iOS and iPadOS 27: The MacStories Review

By Admin Test

After years of relentless building, it’s good to see Apple sweat the details again.

In the 11 years I’ve been writing annual iOS and iPadOS reviews every summer, I don’t think I’ve ever struggled to find an “angle” for my story as much as I did this year. The reason was twofold: I wasn’t particularly enamored with Siri AI and Apple’s belated chatbot efforts; and, as I realized during the summer, it can be challenging to write about an OS update that is largely focused on performance improvements and refinements.

“Everyone wants a so-called Snow Leopard year until it actually happens,” I thought to myself on several occasions over the past few months. Think about it: on paper, we all love a good ol’ basket of bug fixes that just make our phones better without changing much about the overall design and experience of iOS. It’s a commendable effort, and there’s no denying that iOS 27 is the snappiest (insert meme here), most stable version of iOS I’ve ever tested – and it has been since the first developer beta. In practice, though, it turns out that considering a major iOS update merely through the lens of improvements and fixes can be…a little boring.

So I’ve been sitting on this review for a while. I let it simmer. Then, a month ago, as I was sunbathing with my girlfriend Silvia down south in beautiful Vieste, Puglia, it hit me: iOS 27 is a geographically ambiguous update. Stay with me here. Depending on where you are in the world, this update can either be a little boring or pretty exciting.

Apple Watch Series 12 and Apple Watch Ultra 4: The MacStories Overview

By John Voorhees

Source: Apple.

At today’s Surprise and Shine event, Apple introduced the new Apple Watch Series 12 and Apple Watch Ultra 4. Both models build on the devices’ existing designs, adding new health and fitness features as well as what Apple calls Audio Intelligence. The star of the Apple Watch show that powers those features on both models is the brand-new S11 chip. Let’s dig into the details of what the new chip and improved sensors means to both models.

Apple Watch Series 12

The Series 12 takes a heart reat measurement every five seconds. Source: Apple.

The new Apple Watch Series 12 comes in 42 and 46mm sizes and is equipped with the new S11 chip. Apple says that enables the device to take sixty times the heart rate measurements and 24x HRV readings compared to the Series 11 Watch. More heart rate measurements and HRV readings mean more accurate and meaningful readings; in fact, better than any other wearable on the market, according to Apple. Apple says step tracking is more accurate, too.

The Series 12 has improved sensors, too. There’s a larger, more power-efficient green LED array for measuring your heart rate every five seconds and a Secure Exclave for processing audio privately and securely. On the outside, the Series 12 features Ceramic Shield 2, making it tougher than the Series 11. The case is now IP6X rated for dust and can be used to six meters of depth. Plus, the battery is rated to last 24 hours during normal use and a 15-minute charge will power it sufficiently to last 12 hours.

Experiencing Apple’s Live Yankees–Red Sox Immersive Baseball Broadcast

By John Voorhees

Last night, I sat down with my Vision Pro and caught the first part of the Yankees–Red Sox game live in Immersive Video. I wouldn’t call myself a baseball fan these days, but I have fond memories of many summer evenings at Tiger Stadium and Wrigley Field, and I was keen to see how Apple’s immersive broadcast compared.

Having sampled Apple’s other Immersive Video content, but very little of its sports coverage, I wasn’t sure what to expect. Sitting down to watch, I was surprised by how much it felt like I was at Yankee Stadium. Before the first pitch was thrown, the broadcast focused on the Yankees’ starting pitcher as he warmed up. Periodically, the camera angle switched, giving viewers other angles on the pre-game preparations, while the announcers chatted about each team’s season leading up to the game.

Once the game started, the camera switched to positions near each team’s batting circle depending on which team was at bat, occasionally switching to shots behind home plate and in right-center field. During each at bat, a virtual scoreboard showed a 2D feed of the game like you’d see on a regular broadcast bounded by statistics. The additional angle and information were useful for fans, but as someone who was there primarily for the experience, it took away from the immersiveness a little to see a fake scoreboard floating in the sky over the stadium’s right field wall.

That said, the immersiveness was impressive. I found myself looking all around the stadium taking in the sights and spatial audio sounds. The detail was incredible. Fully surrounded by the stadium, I watched as a family walked by with popcorn and drinks and could even read the Poland Spring label on a water bottle in the bullpen.

Then, as I watched the Yankees at bat, I could see someone off camera spitting sunflower seeds in the dirt along the dugout. That sort of minute detail added to the immersiveness of the experience every bit as much as being surrounded by the video feed and soundscape, to the point that when a player entered my view from off camera, I instinctively tucked my legs under me to avoid him tripping over me.

At times, though, I also felt a little too close to the action. It was like I had slipped into a restricted area, an invisible participant, spying on unsuspecting people on the field, waiting for security to discover me and kick me out of the stadium. That feeling was probably driven by the combination of knowing that the broadcast was live and not having watched a lot of Apple immersive sports content, but it was still a little unsettling at first.

As for the game, I didn’t pay very close attention. I don’t follow either team, but it was still fun to watch the game unfold from up close. The dugout camera angles made it slightly harder to see the pitches as they were thrown than it would be from a more traditional outfield camera viewpoint, but it really didn’t take away from the experience.

Ultimately, I only lasted about a half hour. The experience was excellent, but baseball’s pace is just too slow for my taste, so ultimately, I packed up my Vision Pro and caught up on Silo instead.

If you missed the game, it’s available to replay in full along with highlights in the TV app on Vision Pro.

Hands-On with ChatGPT Work’s New Cloud Browser Feature

By John Voorhees

Yesterday, OpenAI introduced a new ChatGPT Work feature designed to let it navigate websites that require a login without revealing your credentials to the model.

Like a lot of people, I spend far too much time clicking around websites that require a login, looking at analytics and other data, checking whether new sponsors have been booked for our podcasts, and more. It’s a tedious but necessary part of my week that slows me down and takes me away from writing and other creative work. ChatGPT’s new Work feature is designed to handle that sort of work for you.

The feature works by signing into websites using a virtual, cloud-based computer. That separates the browsing session from any browsing session on your local computer, walling the agent’s computer use off from your open tabs, cookies, browsing history, passwords, and other data. Because the login and browsing happen in the cloud, that also means work can continue whether you close the ChatGPT app or power down your computer.

According to OpenAI, an additional review model checks the sign-in request and credential destination for signs of phishing or deception. ChatGPT Work then pauses while the user signs in. Credentials entered through its secure sign-in form go directly to the cloud browser and are not visible to the model. After the user authenticates, ChatGPT resumes its work.

OpenAI also explains that its cloud-based browser doesn’t store your username and password. Instead, it saves cookies that allow you to return to a previously authenticated session. If you don’t want to remain logged in, ChatGPT’s cloud browser settings allow you to clear browser data for individual websites or all sites.

Similar to other features in ChatGPT and Codex, users have choices when it comes to which sites ChatGPT Work can access. “Always ask” is the default, requiring the agent to check with the user before using a website, but the permission level can also be set to “Auto approve” after ChatGPT checks a site for relevancy and risk or “Always allow,” which OpenAI discourages. Individual websites can also be allowed or blocked. These settings control website access, but consequential actions require separate confirmation.

ChatGPT Work’s browser control works best on a mobile device and with simple login systems.

I gave ChatGPT’s browser use a try and the results were mixed. I wasn’t able to get the feature to work at all using ChatGPT Work in Safari on my Mac. Some websites had security measures in place that prevented me from logging in. Other times, the cloud browser asked me to take over manually, but I was unable to do so because the UI was frozen or login panels didn’t appear.

Logging into a site with a CAPTCHA required manual intervention.

I had better luck using my iPhone. On one site with a simple username and password system, a login sheet appeared. I entered my credentials, was logged in, and the agent navigated the site to answer my queries. On Apple’s affiliate link dashboard website, I had to take over the browser interaction manually to satisfy a CAPTCHA, which worked after several challenges. Once logged in, the agent pulled and analyzed the data I requested. Also, after I’d signed in to both of those sites on my iPhone, I remained logged in to ChatGPT’s cloud browser, which meant I could also use them from my Mac.

In its current form, the idea of having ChatGPT Work browse signed-in websites on your behalf is better than its implementation. If you can get logged in, having an agent collect and analyze things like analytics data is fantastic. However, getting past the initial login screen is still too frustrating. That said, I’ll be keeping a close eye on the feature, which is available on eligible plans depending on rollout and workspace settings, for future use collecting and analyzing data that would otherwise require a lot of clicking around.

Hands-On with Computer History: OpenAI’s Take on Agent Memory

By John Voorhees

Late yesterday, OpenAI revealed the sort of automation catnip I love: Computer History. It’s a new feature for Pro, Business, and Enterprise subscribers that creates continuity for ChatGPT and Codex by converting the actions you take on your Mac into Markdown memory files that serve as context for subsequent interactions. The idea is that with the additional context, ChatGPT and Codex will better understand your requests, requiring you to provide less detail up front while getting better results. Anyone familiar with OpenClaw, Hermes Agent, and similar projects will know where this is going.

Here’s how it works.

Become a Club Member