If you’re paying $20/month for ChatGPT Plus just to use the API, you’re vastly overpaying.

In November 2024, I used OpenAI’s GPT 4o model via their API 218 times — over seven times per day! — and it cost me exactly $0.62. Now, before your eyes glaze over, using their API isn’t hard. I built a shortcut that does it, and it’s my primary way of interfacing with ChatGPT or AI of any kind.

What do I use the API for? I use their model to generate a description for every image that I post online. For each image that I post to social media and my blog, they all get sent first through to OpenAI to look at, and it returns a short description that I can edit if necessary (it’s rarely necessary). I even built a Make.com automation that takes all the photos I post to Instagram and routes them through OpenAI for a description before automatically crossposting them to my blog.

I used it to generate a description for the screenshot of the chart above in this very blog post. Here’s what OpenAI came up with:

Bar graph displays daily spending in November, with green bars indicating varied amounts. Taller bars appear mid and late month. Total spending is $0.62, text indicates “Monthly Spend.”

I know ChatGPT Plus gets you access to a bunch of other features like more advanced voice usage, more image generation, building personal GPTs, and more. But at $20/month, ChatGPT would be one of my most expensive subscriptions, and honestly, I never feel like I’m lacking by using the free version.

Setting up API usage

I did this a long time ago, so I just had to reacquaint myself with how this works. After logging into your account at platform.openai.com, go to Billing. There, you can set up a Pay As You Go plan. They want you to keep at least $5 on your account, so mine is set to top up to $10 any time it drops below $5. I added $10 back in March and haven’t been charged since. I doubt I’ll make another payment before February of next year. That’ll be a whole year of usage for about $5.

And you never have to worry about something going awry and racking up a huge bill. You can set budget limits. Mine will alert me if I use more than $5 worth in a month, and I’ve set a hard stop at $10.

Once you’ve got a Pay As You Go account, go to the API Keys pane in Settings to create a new secret API key that you’ll use in your projects. You can even create multiple keys if you want to track usage across different projects.

Then get started creating! You can use my shortcut or learn how to do something new on YouTube. OpenAI’s documentation is pretty good, too. And I certainly recommend the meta approach of conversing with ChatGPT to work out your API requests.

Just know it’s cheap to use, and getting cheaper.

Update: I meant to mention that some other apps will let you plug in your personal API key to enable OpenAI-dependent features. Drafts is one example. Some apps will offer that as an alternative to paying extra for AI features since you’re removing them from needing to process your requests through their API at their expense. It can save you from needing to pay each developer directly for AI features, and I’d like to see more apps offered that option.

Nilay Patel, editor-in-chief at The Verge, announcing their new subscription option:

Today we’re launching a Verge subscription that lets you get rid of a bunch of ads, gets you unlimited access to our top-notch reporting and analysis across the site and our killer premium newsletters, and generally lets you support independent tech journalism in a world of sponsored influencer content.

It’s a move that I’ve expected for a long time, years really, and follows behind so many other smaller publications and podcasts that launched memberships when the ad industry tanked during the COVID pandemic. Here’s what they’re restricting for non-paying visitors, and providing for subscribers at $7/month or $50/year¹:

Our original reporting, reviews, and features will be behind a dynamic metered paywall — many of you will never hit the paywall, but if you read us a lot, we’ll ask you to pay. Subscribers will also get full access to both Command Line and Notepad, our two premium newsletters from Alex Heath and Tom Warren, which are packed full of scoops every week.

I’m also delighted to say that subscribing to The Verge delivers a vastly improved ad experience — we’ll get rid of all the chumboxes and third-party programmatic ads, cut down the overall number of ad units, and only fill what’s left with high-quality ads directly sold by Vox Media. It will make the site faster, lighter, and more beautiful — more like the site we envisioned from the start, and something so many of you have asked us to deliver.

I’d prefer no ads, but I’ll reserve judgement until I see some screenshots of this “more beautiful” version. For all that The Verge writers complain about the web becoming unreadable due to the onslaught of ads, their own site has been complicit for years.

It sounds like the vast majority of the stuff I like to read there — well, the vast majority of all their stories — is going to be in the metered group. Plus, I rely heavily on RSS, and it sounds like their feed is going to be truncated going forward.

I question whether putting (nearly) everything behind a paywall will be good in the long run. I’m sure they want their work to be read far and wide. Will turning off the story spigot frustrate their loyal readers enough that they’ll go elsewhere? Or will they be invested enough to cough up the dough? It’s a big bet!

But if the meter is fair (please no “you’ve reached your one story per month” limits), it could work. It could let infrequent visitors read the (ad-overfilled) site as usual, and when a reader turns into a True Fan, they’re charged and given a better experience. And I expect the meter can be fine-tuned if it’s not working.

I’m not opposed to paying for good journalism or content in general. I get enough value from The Verge (it’s the primary place I get my non-Apple, wider tech industry and policy news) that the price seems pretty fair. But I’ll probably start with a month trial to see if it’s that much better.

Oh, and I sure would love for the reduction in ads to apply to The Verge’s podcasts, too. Their ever-lower rank in my queue is a direct correlation to how annoying their ads are alongside their lack of chapters to skip them.

Benjamin Mayo, writing at 9to5Mac on Apple’s latest rendition of the wildly popular personal music trends feature that Spotify debuted in 2016 with Wrapped (but also kind of 2015 with Year in Music):

The Apple Music year-in-review montage for 2024 is now available. View yours now at replay.music.apple.com for Apple Music subscribers. New for this year, in the Apple Music app on iOS 18.1 or later, users can also tap through to watch their Replay without leaving the Music app.

An unreserved finally is due here. For a company that touts the advantage of native apps, it sure took them a long time — they only got their offering together in 2019 — to break Replay out of the web. We all know Apple’s famously strapped for cash and developer talent, though, so who can blame them? 🙃

(But has it really broken out? As far as I can tell, the Music app is just opening a web view of the same stuff available at replay.music.apple.com. It doesn’t really offer anything new or different in the native app. Whatever, I’m calling it a win.)

Here are my top artists for 2024:

Maybe next year Apple will do something more inspired, but for now I’m just happy I don’t need to log into a web app with a decidedly long subdomain just to see some cool graphics. 🤷‍♂️

Someone should probably tell MacRumors though:

For easy access to your most-played tracks, you can add your Apple Music Replay 2024 playlist directly to the Music app. However, unlike Spotify’s Wrapped feature, Apple Music Replay remains a web-based experience, with the Music app only able to play the curated playlist of top songs once added through the website.

😬

Music

I haven’t written a whole lot about politics since the U.S. election in which we, as a nation¹, elected the convicted felon and treasonous Donald Trump back into the presidency.

Suffice it to say that I’m profoundly disappointed in, ashamed of, and embarrassed by the lack of moral character in this country that has led us back here.

But I’ve been hearing a lot of quips from liberals lately that have me worried.

“I guess I need to focus on taking care of me and mine for the next four years; I simply can’t be up in arms about every disaster.”

Isn’t that the behavior that has gotten us here? People neglecting to care for the rest of humanity and instead paying attention only to what affects themselves?

“Democrats need to take a page out of the Republican playbook and say whatever is necessary to get elected, and then just do what they think is right.”

Aren’t Republicans convinced that what they’re doing “is right”? “Whatever is necessary” sure leaves a lot up to interpretation. And when their actions are motivated by hate and fear, we have seen firsthand that the results are disastrous.

I recognize that these things are said out of a sense of defeat. What we’ve tried obviously isn’t working, so we need a new approach.

Right on cue, here’s Manton Reece, writing today after President Biden pardoned his son, with a subtly different sentiment:

Democrats need to rethink the old rules. Democrats need to say and do what’s right without giving a fuck what anyone else thinks. This pardon might be the first step.

My initial reaction is to pump my fist and give an enthusiastic “Hell yeah!” to what Manton proposes. To jump aboard the idea that we need to stop overthinking, gnashing our teeth, and fretting over every decision. Good on Biden for taking executive action where he sees fit. It’s what we elected him to do.

“Just do what’s right” seems like a simple and solid approach. Who could argue with it?

But I fear this “throw caution to the wind” attitude can lead us down a dark path. A slippery slope. The same one that MAGA/Republicans/Conservative/right-wing folks travel. One where we, like them, act out of self-interest rather than public interest. That we let fear and hatred of those across the aisle motivate us rather than love, kindness, and compassion for our fellow citizens of Earth.

We already call each other “monsters”. Villify each other. Refuse to see the humanity in each other. Do we really want to add another similarity to that list? That we, too, would say whatever it takes to get elected and then abandon any campaign promises to act solely on our gut?

I want to believe that if we all just do what’s right, we’ll make our way out of this mess to a better future. But without some consensus on “what’s right”, I fear we all start acting out of self-interest. And I just don’t believe more people acting in self-interest is going to do more good for the world.²

It’s late, and this turned more philosophical than I intended. I don’t feel like I’ve come to a strong conclusion, either here or within myself. More thought is necessary.

I doubt I’ll switch to The Browser Company’s new AI-infused-and-focused Dia browser when it debuts next year. But I can’t say I’m not impressed with their latest video. The meta angle of explaining the structure of this recruiting video within the recruiting video was a creative move. Their whole company is quite creative, really. Their Arc browser is nothing if not a bunch of creative new ideas about the interface for a web browser. One that didn’t click with me, but I can see why it clicks with so many other people.

I also think they’re dead-on when they say that AI isn’t going to be a bunch of different apps or buttons within every app. It’s going to evolve into a layer on top of everything we do with our devices. But I think that layer is going to be most useful at the operating system level, not contained to the web browser.

Still, if you don’t have an operating system to play with — and there are only a few in the world that matter — the web browser is the next best environment. While it doesn’t have access to all the other apps and data on your device (which is why plugging into the OS’s AI system is going to be most effective), a whole bunch of what people do on their computers is done in websites and web apps in the browser. And that percentage is growing, not shrinking.

Putting generative text features behind the text cursor is a genius move. I hope more products jump onto that UI. It now seems so obvious that the cursor (caret) is exactly where those capabilities should live.

The browse-for-me/take-action-for-me approach is a familiar idea. It’s what the Rabbit R1 is supposed to do. On one hand, I like the approach because it means interfaces can continue to be built and optimized for human use, and the computer just needs to learn how to work its way around the human interface. On the other hand, it seems pretty inefficient when the alternative is for sites to provide APIs for the computer to tap directly into, rather than fake its way around an interface.

I don’t know which is going to win out there, but building the ability to “click around” a human interface will always be a good backup for sites and services that don’t provide APIs.

The Browser Company is certainly one to watch in our next phase of computing. They’ve already earned a loyal fan base with Arc. That’s the toe in the door to becoming a more pervasive influence in the tech world.

Those fans — and others like me — will try out Dia just because of their reputation. If it’s great, we’ll tell our friends. Enough people switch from the big browsers, and there’s some real disruption to the current computing landscape.

Especially now, in this tumultuous time when Google has been directed by the Department of Justice to sell off its Chrome browser, The Browser Company seems poised to be a bigger player in the next chapter of the web.

A weekly list of interesting things I found on the internet, posted on Sundays. Sometimes themed, often not.

1️⃣ I think you’ll know where this video is going once you see the first few seconds, but see it through. Like most things from Cabel Sasser, the payoff is good. [🔗 Cabel Sasser // cabel.com]

2️⃣ Want to wear a literal emoji? Now you can because Jose Wong produced a shoe based on Apple’s sneaker emoji. 👟 [🔗 Andrew Liszewski // theverge.com]

3️⃣ 365 marathons in 365 days. £1 million raised for hospice care. It’s a happy tear jerker. [🔗 Great Big Story // youtube.com]

4️⃣ Far too many good quotes here about how running helps writing to pick just one. [🔗 Brendan // semi-rad.com]

5️⃣ I wouldn’t mind getting guided on a tour by a Boston Dynamics Spot robot powered by an LLM. [🔗 Boston Dynamics // youtube.com]

6️⃣ DMVs aren’t always bad places. This one was affirming for Lisa Melton. [🔗 Lisa Melton // lisamelton.net]

7️⃣ Simon B. Støvring has made a little utility for himself to decorate his macOS dock with holiday lights. I, for one, hope this gets released to the public someday. It’s so festive! [🔗 @simonbs // mastodon.social]

Thanks for reading 7 Things. If you enjoyed these links or have something neat to share, please let me know. And remember that you can get more links to internet nuggets that I’m finding every day by following me @jarrod on the social web.

7 Things

A weekly list of interesting things I found on the internet. Sometimes themed, often not.

1️⃣ A great reminder from Rick Rubin, via Steve Ledlow. [🔗 tangiblelife.net]

2️⃣ Loved Fleabag. Definitely love it in Lego. [🔗 Trevor Carlee // youtube.com]

3️⃣ Follow this Mastodon account and it’ll automatically reply to all your posts with an image with an alt-text description of it, making them more accessible to folks who use screen readers. [🔗 @altbot // fuzzies.wtf]

4️⃣ @heymikli on Threads has an absolutely stacked shortcut for their Action Button to do different things based on the phone’s orientation. I’m gonna try it out. [🔗 threads.net]

5️⃣ This guy seems awesome. I hope I see him climbing at the Gunks someday. [🔗 Eric Bates // youtube.com]

6️⃣ This example site renders everything in a monospaced font grid. It’s beautiful. [🔗 Oskar Wickström // owickstrom.github.io]

7️⃣ I’m super diggin’ this new (to me) podcast. Super short, super interesting little stories. The first episodes are kinda macabre, but fascinating nonetheless. (Via 99% Invisible) [🔗 the memory palace // overcast.fm]

Thanks for reading 7 Things. If you enjoyed these links or have something neat to share, please let me know. And remember that you can get more links to internet nuggets that I’m finding every day by following me @jarrod on the social web.

7 Things

In all the hubbub back in September wondering how Apple was going to advertise its new iPhones with Apple Intelligence when Apple Intelligence wasn’t even shipping on the new iPhones yet, I don’t recall anyone speculating how Apple Intelligence itself would be advertised. Well, we have our answer, and it’s not great.

From the first Apple Intelligence ad, something didn’t sit right with me. Let’s watch it together:

Did you catch it? No? How about another:

What about that time? Do you need another? Here:

Clearly, you caught it that time, right, especially since my lede is in the title. I think this is the first time Apple has actively encouraged us to lie to one another in one of its ads.

Worse yet, it’s all so avoidable. You don’t have to deceive your friends and coworkers. Be honest with them and say you forgot to read the email, that you remember your last meeting but not their name, or that you need a little time to catch up on the prospective. Then use Apple Intelligence to get your work done faster. Don’t pretend that you know something that you don’t, or did something that you didn’t. Don’t do whatever the hell this guy is doing. And, for crying out loud, don’t be so fucking awkward about using your computer as that guy did in the last one.

LLMs (and AI by extension) have been called “bullshit machines” because they spout out anything and everything they think will next most likely word in response to your query. Not because it “knows” the right answer, it’s just going to bullshit its way to an answer. Fake it till you make it.

Apple Intelligence isn’t all that different in that regard. I expected its LLM-powered features to bullshit me. I didn’t think they would shown off, by Apple as enabled by Apple Intelligence, to deceive one another about the status of a project or our memory of our last meeting together. Almost everything they’re showing off about Apple Intelligence is a way to make it seem like you’re more thoughtful, attentive, and prepared than you really are.

And that just doesn’t seem right to me. Show a parent and child imagining a story together, with Writing Tools helping them to adapt the story to the child’s every fanciful whim, and Image Playground bringing it to life with otherworldly illustrations. Show people building study quizzes with AI based on notes they’ve taken and lecture transcriptions. Show movie memories being made with unlikely themes, generated from your photos and videos based on an imaginative prompt (but not thrown together at the last second and then passed off as a thoughtful gift).

In fact, this is probably the only good Apple Intelligence ad I’ve seen so far:

The guy was able to get his frustrated emotions out by typing an unhinged email. He used Writing Tools to tone it down so that he wouldn’t be an asshole to his coworker. They even showed him perhaps learning how using kind words both got him his pudding back and earned him the respect of the thief. All good stuff, humorous, and legitimately useful. Who among us hasn’t typed out a nasty-gram only to delete it and then have to write more professionally? Apple Intelligence can help in that situation, saving us time, just as computers have always been designed to do.

I’m not the only one who has noticed this disheartening new direction for Apple’s marketing. Here’s Anil Dash on Threads:

It’s not important at all, but it’s interesting to note that Apple has made another recent ad that’s as bad as the hydraulic press one where they destroyed all the instruments. In this one, it shows their AI tool being used by someone who didn’t do their work to fake their way through a meeting. Apple ads used to always show their users as experts or creative thinkers. Now they’re workplace liars.

And hapax, who wrote so succinctly about this that I’m going to quote them in full:

My wife and I have been watching Parks and Rec a lot lately. Every ad break it seems like one of Apple’s ads for “Apple Intelligence” comes up. These ads feature someone suddenly realizing that they’re unprepared: a wife realizes that she forgot her husband’s birthday; a lawyer realizes that he forgot to read an important document; an actor realizes that they forgot to read a pitch for a show. The unprepared person is able to quickly “save the day” by using Apple Intelligence to generate a summary of important information, or in the wife’s case, to generate an animated slideshow of sentimental photos.

These ads consistently leave me with a bad taste in my mouth.

Obviously, everyone is unprepared sometimes. Everyone occasionally forgets an important birthday, or shows up to a meeting without having prepped. That’s part of being human. The myth of AI is that we can eliminate human fallibility, that we can smooth out our rough edges with enough technology. News flash: more technology won’t make us more perfect. The more we expect “perfection” from others, the less grace we will have with them. The less space we leave for forgiveness and understanding.

Show us creating something new, more, and better with Siri, empowered by Apple Intelligence, as our helpful assistant. Don’t encourage us, in a time when we’re already far too mistrustful of one another, to deceive our friends, family, and colleagues with the click of a button.

One of my favorite videos that Apple has ever put out was this one.¹ In it, they profess, “Give people wonderful tools and they’ll do wonderful things.” It’s a sentiment that I took to heart, and that I truly think they, as a company, believed at the time. Today, a version of that video might instead read, “Give people bullshitting tools and they’ll do bullshit things.” Come on, Apple, do better.

Andrew Liszewski, theverge.com:

An Italian startup called Artinoise has created an unusual USB-C accessory for mobile devices that turns them into playable musical instruments. The Zefiro looks like a flash drive or a tiny vape, but by gently blowing into one end, it can be used to play simulated instruments with even less skill than what was needed to play those plastic recorders in grade school.

As soon as I saw that hero image, something clicked in my memory hole and I frantically searched back through my App Store history. Yes! I found it! Purchased back on December 31, 2008, Ocarina by Smule.

This app — launched very early on the App Store — explored using the iPhone’s combination of hardware capabilities to create something unique: a smartphone instrument. You blew into the microphone on the bottom of the iPhone while covering combinations of four “holes” on the screen to play different notes. It was one of the coolest demos you could give your friends on an iPhone at the time (besides the iBeer app, of course).

The Zefiro looks like it uses a similar interface, 16 years later, but instead of blowing into the microphone, they’ve made a plug-in mouthpiece. Admittedly, this is an improvement. I remember always feeling slightly like I was hyperventilating while playing the Ocarina app, seeing as there was no resistance when blowing just through my lips.

Here’s the best thing, though, and I’ve buried the lede here. Ocarina lives! I think you can still download it new today, but I was able to download Ocarina from my Purchases screen on the App Store. With bated breath, I tapped the ‘Open’ button once it had downloaded…and it works perfectly on my iPhone 16! I ran through the tutorial and, with surprising ease (which means I must have really built that muscle memory back in High School), played ‘Twinkle Twinkle Little Star’ on my little smartphone instrument.

It really made my day. I hope you can try it out too. If not, here’s a very of-the-time YouTube video of how it worked.

Apps

A weekly list of interesting things I found on the internet, posted on Sundays. Sometimes themed, often not.

1️⃣ What The Verge did last month to look back at the internet’s infancy in 2004 was remarkably fun. Really, go get lost in this time machine of special project. [🔗 theverge.com]

2️⃣ I don’t know if this is true about CBC’s sound signature, but I’d love it if it is. [🔗 @nickheer // c.im]

3️⃣ Russia has fined Google an amount of money more than all there is on Earth. Uh, okay. [🔗 Joshua Nelken-Zitser // businessinsider.com]

4️⃣ Boston Dynamics videos never disappoint. Their fully electric (not hydraulic) and fully autonomous robots are hella impressive. [🔗 Boston Dynamics // youtube.com]

5️⃣ Arun dives deep into the Leica camera that Jony Ive and Marc Newson redesigned a number of years ago. Still one of the prettiest cameras I’ve ever seen. [🔗 arun.is]

And two Mac mini things to celebrate the arrival of its svelte new shape:

6️⃣ Oh boy do I ever want this Mac mini Pro enclosure! [🔗 @basicappleguy // threads.net]

7️⃣ Worried about the Mac mini’s bottom button? Here’s a fun solution that turns the whole computer into a power button, Easy Button-style. 3D-printing files coming very soon, it seems. (And the thing appears to suction the Mac to the desk, which is probably good considering how lightweight it is now.) [🔗 gijsmans3773 // reddit.com]

Thanks for reading 7 Things. If you enjoyed these links or have something neat to share, please let me know. And remember that you can get more links to internet nuggets that I’m finding every day by following me @jarrod on the social web.

7 Things