Loading summary
A
Anthropic has just launched a brand new model, Claude Opus 5. It is half the price of Fable 5, but almost as good as Fable 5. This, I think, is in direct response to what OpenAI has been doing. And speaking of which, OpenAI has just brought ChatGPT Voice to their desktop app and it has agent control. So talking to it, it can then control your computer. Meta is pivoting their AI chatbot towards productivity. With Musespark 1.1. We'll get into what they're doing there. But hugging face Meta, Nvidia have all signed, along with over 200 Silicon Valley companies, a letter against open weight AI restrictions and researchers using AlphaFold Fold to redesign CRISPR Cas9 and cut off target edits from 28% to 5%. MidJourney has just acquired Costars. It's an astrology app. It has 4.3 million monthly users and it has nothing to do with Mid Journey, but the talent that have built it have everything to do with their company. So we're going to get into that. But first, my absolute favorite AI tool I've been using this week allows you to generate images and audio and video right inside of Claude, who, you know, only does text and reason. And I basically use Claude as my control panel for everything I do. I have two Claude Max $200 a month subscriptions and I burn through all of the tokens for both of those because I have a billion projects going all, always all at once, and this thing is a workhorse. Now the problem with it is it doesn't generate images like ChatGPT, it doesn't generate video like Google, and it doesn't do audio like 11 labs. So I've actually been able to pull all of those models straight into Claude with the AI box mcp. It's amazing. I've created tons of Facebook ads this week. And the thing that actually saved me a ton of money and also just time is I was creating a bunch of podcast covers for a project where it was a. It was a language learning podcast. So it's like, learn German, right? And it'll be a podcast in German. But my problem was I was like, you know, people that are from, you know, Spain or from Russia that want to learn German, they're not going to search for learn German in English or even learn German, and however you say it in Dutch, they're going to search it in their own language of Russian or whatever. So anyways, I have this problem and the solution is generating podcast covers with ChatGPT. You know, learn, you'd say the words, learn German, and you'd Translate it to 10 different languages. Okay. So I had this great podcast cover design that I really like, and I was just getting chatgpt. I was giving it to it and it was spitting out a new title. And I know a lot of people are like, oh, why don't you just use Canva to do this? Yeah, you could use canva to do it. And then you'd have to manually adjust, like, the font sizes because once you get into, like, Mandarin or Korean and like, the languages are completely different. I had two different fonts and two different colors. And anyway, I was gonna have to go Finnick with it and going to be annoying. So I was having Chat GPT just one shot, this, but it was taking forever, and then this. For this particular project, I'm actually doing every language and every language is variant, right? So it'd be like, learn English, learn German, learn French. They're podcasts in those languages. But it's. I'm. I'm making a podcast cover for every language, right? So there'd be like 10 or 12 variations of everything. Anyways, it was a big project. It was going to take Forever to generate 300 podcast covers for every language variant. And so what I did is I had Claude make, you know, the base image for every single one of these, which is like 12 languages. So it just made the base image for the 12 languages, and then Claude actually put the text on and it can swap it out. It saved me so much time being able to use Claude +AI box. If I was just going to chat GPT and trying to get, you know, 300 images generated, this would take me forever. But, but getting Claude to. Basically, I had the AI box MCP called in. So it just said, hey, you know, Claude knew exactly what to do. It wrote the prompt, I gave it an image and told it what I wanted it to do. It wrote the prompt, it used AI Box, it pulled the image in, it put the text on, and swapped it out into all the different languages and made the template perfect. So I got to use the brain power of Claude, but I got to use the image generation power of chat GPT. Using them together, I only had to generate 12 images. And I saved myself so much time and money being able to do it this way. There's like a hundred other things I've done this week with it. That was just like a fun thing I was doing last night with it. Testing out. Go try out the AI box mcp, if you haven't already. It can generate images, audio, video, right inside of Claude. You use Claude's brain. And by the way, it's not just Claude. Like, you can do the same thing inside of Codex or inside of Cursor or inside of any of these other tools that pull MCPS in. Okay, let's talk about what's going on with Opus 5. This one's actually a pretty fun model, but because the biggest issue that I've seen with Claude at this point is that Fable 5, which I use, I burn through credit super fast, but it also burns through all of my tokens for my entire account. So the new Opus 5 is very similar on the benchmarks. It's slightly worse than Fable 5, but very close, and it is half the cost. This is amazing. I will take a, you know, a 10% or a 5% cut on, you know, brain power quote, unquote, you know, in order to get half the cost down. Because I'm just burning through so many tokens using it. I only use it for, like, the most important, critical tasks. So I think from now on, I probably just won't use Fable 5. I'll just be using Opus 5 for everything. And I think this is in direct response to ChatGPT, who came out with a very similar model. ChatGPT's latest, I think, 5.6 is half the use about half the tokens that Fable 5 was using. So Claude obviously wanted to come up with a competitor because I was, you know, like, basically once my subscription to Claude runs out in August, I'm switching over to ChatGPT, into Codex and using ChatGPT work on my desktop for, like, coding and everything, because of all the features they have. But in the meantime, I'm still using Claude till that runs out. But anyways, I think Claude knows this, and so they. They're making. Basically, they're trying to get themselves on par with OpenAI, come up with all the features that were just launched. And speaking of really cool features from OpenAI, OpenAI actually added chat GPT voice to the desktop app, if you remember. I don't know if you, you know, everyone's at a different phase, but basically, OpenAI has a desktop app called Chat GPT. They used to have one called Atlas, which was like a browser, and they had one called Codex, which was for coding. They combined them all. It's just called the Chat GPT app. And they now have Chat GPT voice that they've added to it. Which what that means is it lets you talk to Chat GPT to this thing that controls your Computer, right? It does everything that Claude Cowork can do where it like, controls your computer, it can control your files, it can open your browser, can control your browser, it can do all that kind of stuff. But while you're talking to it, so you can have something pulled up on your computer screen and you could be writing code instead of like, I do this all the time with Claude. I go take a screenshot of whatever I'm working on and then I, you know, drop it into Claude and I'm like, what do you think about? Blah, blah, blah. Okay, now with ChatGPT Voice, you can have something pulled up on your computer screen and you could just be talking to it. You're like, hey, you know, I'm on my emails and it looks like they said this thing is missing, or I'm in my files or I'm in my software and there's this issue and it's looking at it and it can then debug it and test it right from there. I am so stoked about this. You have no idea. Like, this is actually massive. You're no, it. It's just getting to the point where whatever's on your computer screen is. You're going to be talking to it about that in particular. So, yeah, this is a big upgrade. I. On Mac os, there is a feature called App Shots and it basically lets voice agents see what's on your screen, including the alt text. So what I love about it being able to see the alt text. These agents are really cool and I use Claude for some of these similar types of tasks a lot where beyond just seeing what's on your screen. So, like, I have a project where I have to download 300 hours of podcast video content from the platform that I hosted, all on Riverside. I've been recording there for years. I wanted to export all my podcasts off of there and I, you know, I like, email their support and I'm like, hey, do you guys have a bulk, like, export thing? Whatever, I'll pay extra. They're like, nope, sorry, there's nothing we can do. No bulk export. You got to go manually, click every single one, and you only can do 15 hours a month. And I'm like, I have 300 hours. What do you mean? Anyways, so there was like a couple little loopholes they came up with. It was super annoying. Obviously I'm not going to spend a week of my life manually, like, exporting every single thing. So I got Claude to go look at it and when Claude went in there, it was like, hey, look, we actually found an API inside of this, and we're able to, like, automate and manual or, you know, automate a lot of the stuff and download everything we found, like the download keys and the files and whatever. So Claude's in there doing who knows what. But it's not just like looking at my computer screen and clicking on actual buttons. It's like executing code to go and do a lot of this stuff programmatically, which is really cool. And I mean, basically that's what this new Mac OS feature is able to do beyond just looking at what's on your screen and looks at what the code is, the alt text, the code, what's behind the actual screen. And so I'm like, so stoked about this. You've probably heard me talk about this, or you probably use this yourself, where, you know, these. These coding models are not just looking at the visual, they're looking at the code behind everything. I was, I think, like a couple months ago, I was asking Claude to edit a video on Cap Cut. It closed my Cap Cut app, opened the Cap Cut file, made some tweaks inside the code, and was like, okay, your video's edited. And I opened it and it, like, had cut it up the way I wanted. So these, these models are awesome. And they often will just look at the code and not even do it the visual way. So really stoked to use that with OpenAI, Meta is pivoting their AI chatbot from just being kind of like entertainment. They're trying to focus more productivity. In the past, you've probably seen all the branding with Meta where it was like, chat with the Mr. Beast AI on WhatsApp and like, go generate an image of your friends, generate a picture of you as a pirate, right? Like, this is kind of like the fun viral stuff that they were working on. But I think at the end of the day, they're realizing people are not going to pay for that. Like, it's fun and it's a novelty, and maybe it was cool, like, I mean, years ago, but it's just not really driving actual usage today. And so they're now moving more towards productivity, more in line with what Claude and what ChatGPT are doing. They're adding calendar access, they're adding daily briefings, they're adding recurring task automations that directly compete with, like, I have task automations on Claude that will run every week or every day or some of them every three hours, depending what the task is. So I think we're seeing a big shift for Meta. They're getting away from just being fun and they're going to productivity and I think we're going to see them get a big spike in, in usage. Meta can also now read your calendar. It can plan events, it can look at Facebook Marketplace for items within a certain budget, which is kind of funny because I remember ChatGPT was doing, doing this for a lot of people with ChatGPT agents. They were like getting it to lowball people in Marketplace. But anyways, I guess you can now do that with metazone Agent and this new update, it's going to be out today in a bunch of different markets and WhatsApp and more regions are going to follow in the next couple weeks. Meta has some massive distribution advantages that not, you know, most tech startups do not have. They have WhatsApp, which reaches 3 billion users, and then they also have Meta AI and that's already built into Instagram. They have messenger, they have Facebook, so they have a lot of distribution. Let's see if they can stick the landing on this hugging face. Meta and Microsoft, Mistral and Nvidia are all signing an open letter and I think There was like 200 other companies that did, urging the White House not to restrict open weight AI models or ban the technique of training one model on another model's outputs. Model distillation. This is so fascinating because there's a, you know, this letter in particular, they're pushing back on a plan to respond to Chinese AI labs by penalizing open source development alongside actual IP theft. Now this is coming from. I was reading the tweet earlier today from Secretary Besset from the treasury and he basically came up with this tweet, which, I mean, I don't know anything about the government, but I don't know why our Treasury Secretary is weighing in on open source AI models, but whatever. And basically he was saying, hey, look, the Chinese labs are doing model distillation, asking, you know, Claude to give them like 5,000 answers to a question and they use those 5,000 answers to help train their model. Now a lot of open source models are using the same tactic. Some people think it's unethical in my opinion. I just think it's so ironic because anthropic just spent $20 million. They paid another $20 million to a group that is kind of fighting against this. Anthropic is constantly saying, hey, look, we're being like model distilled by Chinese labs and they come up with these big reports and like, yeah, I don't want to just spend all of our money and then give the, give the secret sauce to the Chinese models for free. But at the same time, I think it's pretty hard to get around that. And it was actually interesting. Elon Musk, when he was in a lawsuit with OpenAI, they asked if he'd used if Grok had used model distillation from OpenAI to train their model. And he's like, yeah, everyone does that now. I know, like a lot of people are up in arms about this or whatever and they're like, look, no, like we gotta fight back against the Chinese labs or whatever from ripping off our IP K. At the end of the day though, I just find this Incredibly ironic that OpenAI, Google and Anthropic have scraped the entire Internet of every blog and article ever written. They've made these models that are incredibly useful, get, you know, a trillion dollars in values being created by these things and then which, you know, they scraped for free. Like everyone wrote blog posts, they scraped the blog posts and they trained them into here. And then once they do that, they're like, okay, no one else is allowed to distill from our model, which is distilling from the human corpus of information, which is the Internet. So anyways, I actually am much less sympathetic towards this, so I would sort of side with Hugging Face, Meta, Microsoft and like half of Silicon Valley, it's kind of split. I think companies with these AI models that have spent billions training them, they don't want you to be able to do it. But for the open source models, I mean, this is basically how you make these open source models so good. And then there's just the geopolitical thing because Kimi K3 coming out of Moonshot, apparently copied Anthropic's Fable model to build their model, which. Interesting, right? It's not even like they were going after, they were islanding like Opus. They're like, no, we're going after Fable 5 to train this thing smart on them. But that model is really good. Kimmy K3 hugging faces use is a Chinese open weight model which is Zai's GLM 5.2. And I've actually used some of Zai's models. They're good. What's interesting about that particular use case though is that OpenAI actually created a model and this was kind of yesterday's news, but it they created a model, it escaped containment in. Supposed to be in a sandbox, not able to access the Internet, escaped containment, got out and went and was hacking hugging face, then hugging face while their website was being Hacked. They tried to use OpenAI's model to stop this and the guardrails on OpenAI's model actually prevented them from being able to stop the attack. So Hugging Face actually went to the Open Weight Model GLM 5.2 and they use that to defend against that cyber attack from OpenAI. And OpenAI came out and was like, wow, yeah, this is crazy. Our thing got out of containment and it was like hacking you and like whatever. Anyways, all this to say there is no way in my mind that we are trusting a handful of closed source models to be the only one that can produce models. Let out the open source models. Look, it just, it literally just stopped a hacking attack escaping from OpenAI. If, if it's a counterweight to what's going on, in my opinion, we have to have this. But anyways, the Trump administration I think is being lobbied by OpenAI and Anthropic to shut these things down. They're dangerous, yada yada. I think we need open source models to run the Internet. So I am on the other side of that. Researchers have just used AlphaFold to redesign CRISPR Cas9 to cut harmful off target edits from 28% down to 5 percent1 while keeping the gene editor's accuracy on intended targets intact. So this is kind of a science one for you, but I'm excited about this because basically off target cuts are things that will damage healthy genes in patients. So they want to do some of this CRISPR gene editing stuff. But if you have these off cuts, it's really bad for you. So they use AlphaFold to cut that down, which is really cool. It's basically a big step in safer gene therapies, which I'm excited about. There was a Chinese team that tested 23amino acid swaps at 10 key positions and they published the results in Nature today about this. And they found that over 95% of the off target sites changed which amino acids inside of CAS9 touched the guide RNA even when the proteins overall shape stayed the same. So you know, they've, they've made some massive improvements there. I'm really excited about it. Midjourney. This is a company that's been generating images for a long time, but it's kind of stuck inside of Discord. And I guess they do have their own website, you can access them. I'm gonna be honest, I have not used Midjourney. They might be the best image model in the world, I'm not sure. Last thing I remembered is they signed a deal with Meta to get embedded to power their image models. But it's really hard to use them because they're just inside of Discord and on their own website they don't have like an app. So anyways, there's just less surfaces for them to touch. I think they've been struggling because of that. And their response to this is that they've actually went and acquired a company. It's called Costar. It is a social astrology app with 43 million monthly users. It has nothing to do with Mid Journey, but what's interesting about this is that it's pulling in 24 consumer app builders into Mid Journey. So in my opinion, the reason why they're buying this, they're like, look, we need to build an app. Let's go buy a successful app and bring all of their developers in and they can help build the Mid Journey app for us. And, you know, they won't be just a Discord only tool. Costar, as it currently sits, it just generates horoscopes using AI and they say also human writers. I'm sure AI does a ton of this stuff. Then it lets users share astrology charts with their friends. There's like some social features and horoscope stuff that's in there. But regardless, Mid Journey basically has no iOS or Android app. And, you know, OpenAI, Google, anthropic, like everyone has one, every major tool. And so I think Mid Journey right now is building. Is. Is. Yeah, basically trying to build the team that does this. It's also interesting because Mid Journey feels like they're a little bit all over the place. Maybe not super, but they also have like a medical division where they're building like physical spa concepts that scan your whole body and tell you if you have any issues. So they're. They're working on a couple different things. But I would say this astrology app isn't just like a new pivot or direction for them. I'm pretty sure this astrology app is. They're grabbing 24 talented people and they're going to build out impressive Mid Journey app, which I'm excited for. And by the way, Mid Journey would be an incredible company if they just built an API AI box. My company where I, you know, I have over 80 different AI models inside of there. We don't have AI or we don't have Mid Journey on there, which is a huge bummer, but it's just because they don't have an API and people have made like, unofficial APIs that go and like make an account and they're like, kind of scraping discord, but like, unofficial APIs could get your account banned. It could crash. Anytime some UI thing or something changes with Mid Journey, they'll stop working. It's just not reliable. We can't use it is probably against their terms. So anyways, if they made an API, we'd include them and they would get access to, you know, thousands more users from us and eventually millions more users from the overall development community. So I think that would probably be the biggest thing they could do. We'll see what happens with a Mid Journey API. Guys, thank you so much for tuning into the podcast. If you want to get images, audio, video inside of Claude, inside of whatever AI tools you're using, 80 different ones, it's. Instead of paying $20 subscriptions to every platform, go and get an $8.99 subscription to AI Box AI. You get every single AI model and it plugs into Claude and everything else with an mcp. So talk to Claude, an image pops up inside of there. You can literally see the image. You can edit it in there. I use it all the time for making websites where I'll, you know, every website looks like AI slop. And I'll be like, hey, go use AI Box to generate custom graphics for this website. And it makes it look beautiful. Guys, I'm telling you, just use this. This is actually going to be the most useful AI tool you got. It's AI Box AI. If you want to see what we're doing with the mcp, it's Slash mcp. I'll leave a link in the description, but hope you guys all have a fantastic rest of your day.
Podcast: How I AI Stuff
Host: How I AI Stuff
Episode: Navigating Opus 5 and Voice Innovations
Date: July 24, 2026
This episode dives deep into the latest developments in artificial intelligence, spotlighting major product launches, strategic company moves, and a fast-changing debate over open versus closed AI models. The host unpacks key advances such as Anthropic’s release of Claude Opus 5, OpenAI’s new ChatGPT Voice with computer agent controls, Meta’s pivot towards productivity with its Musespark 1.1 update, industry resistance to open-weight AI restrictions, breakthroughs in gene editing with AlphaFold, and strategic acquisitions by MidJourney.
“I got to use the brain power of Claude, but I got to use the image generation power of ChatGPT. Using them together, I only had to generate 12 images.” (05:01)
“I will take a 10% or 5% cut on brain power ‘quote-unquote’ in order to get half the cost down... I probably just won’t use Fable 5. I’ll just be using Opus 5 for everything.” (08:05)
“Now with ChatGPT Voice, you can have something pulled up on your computer screen and you could just be talking to it... it can then debug it and test it right from there. I am so stoked about this. You have no idea. This is actually massive.” (11:04)
“They’re realizing people are not going to pay for that [novelty features]... They’re getting away from just being fun and they’re going to productivity and I think we’re going to see a big spike in usage.” (16:07)
“It is incredibly ironic that OpenAI, Google and Anthropic have scraped the entire Internet... and then once they do that, they’re like, okay, no one else is allowed to distill from our model.” (21:05)
“We need open source models to run the Internet. So I am on the other side of that.” (25:32)
“This is a big step in safer gene therapies... they’ve made some massive improvements. I’m really excited about it.” (27:51)
“I’m pretty sure this astrology app is... grabbing 24 talented people and they’re going to build out impressive Mid Journey app, which I’m excited for.” (30:17)
On Collaborating AI Models (Claude + ChatGPT):
“If I was just going to ChatGPT and trying to get 300 images generated, this would take me forever. But getting Claude to... write the prompt, use AI Box, pull the image in, put the text on, and swap it out into all the different languages... saved me so much time.” (04:58)
On AI’s Expanding Control Over Computers:
“It’s just getting to the point where whatever’s on your computer screen is—you’re going to be talking to it about that in particular.” (11:26)
On the Open-vs.-Closed AI Model Debate:
“At the end of the day though, I just find this Incredibly ironic... everyone wrote blog posts, they scraped the blog posts, and then once they do that, they're like, okay, no one else is allowed to distill from our model...” (21:05)
“There is no way in my mind that we are trusting a handful of closed source models to be the only one that can produce models. Let out the open source models.” (24:38)
On AI Security Incident:
“OpenAI created a model... it escaped containment... went and was hacking hugging face... Hugging Face actually went to the Open Weight Model GLM 5.2 and they used that to defend against that cyber attack from OpenAI.” (23:12)
This episode provides a comprehensive look at the latest advancements and debates in AI, blending the host’s personal tech workflow with sharp analysis on competitive moves, technical innovations, and policy controversies. Listeners gain both a sense of where the industry is heading and practical tips on leveraging state-of-the-art AI tools in their own projects.