Resources:
Join the discussion: Got something to say? Let us know on LinkedIn and network with other AI leaders
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson: LinkedIn Profile
Start Here Series in our Inner Circle Community: Join for free access
AI Industry News: OpenAI and Anthropic’s Competitive Moves Reshape Enterprise Strategy
This week, several pivotal developments in artificial intelligence signaled meaningful shifts for enterprise adoption, workflow automation, and competitive positioning among major AI providers. Business owners and decision-makers can draw direct implications for technology strategy, procurement, and workforce planning from these events.
OpenAI Frontier Platform: Enterprise AI Governance and Integration
OpenAI introduced the Frontier platform, designed specifically for large enterprises to build, deploy, and manage AI agents from OpenAI and third-party vendors [00:02:49 - 00:04:31]. Unlike consumer-facing tools, Frontier focuses on solving "agent sprawl" by streamlining fragmented AI toolsets, disconnected workflows, and siloed data lakes inherent in complex organizations.
Key features:
Unique agent identities: Each AI agent receives distinct permissions and guardrails to secure regulated industries and support governance.
Cross-environment operations: Supports deployment across local clouds, on-prem environments, or OpenAI-hosted infrastructure.
Early enterprise access: Companies such as Intuit, HP, Oracle, and Uber have already contributed to the platform’s development, indicating real-world validation and prioritization of big enterprise requirements.
The anticipated downstream effect: solutions developed within Frontier may eventually influence mainstream OpenAI products like ChatGPT and Agent Builder, making advances in enterprise compliance and agent management accessible to smaller organizations.
Anthropic Claude Plugins: Threats to SaaS in Finance and Legal
Anthropic’s release of new plugins tailored for its Claude CoWork platform made immediate waves in the legal and finance sectors [00:10:41 - 00:14:42]. These plugins automate high-touch tasks such as file management, document drafting, and folder organization, positioning Claude as an alternative to established software-as-a-service (SaaS) solutions.
Market reaction was dramatic:
Software ETF Down 6%: The leading software ETF registered its steepest single-day drop since April following the release.
Major Vendors Impacted: Thomson Reuters stock dropped 15%, and LegalZoom tumbled 20% as investors reacted to the prospect of AI-native products displacing expensive legacy solutions.
Investors and analysts are closely watching for further disruption as professional services automation matures, with particular risk for companies reliant on “billable hour" models and licensed SaaS platforms.
AI Advertising: Anthropic vs. OpenAI and the Business Model Debate
Amid the Super Bowl advertising blitz, Anthropic took a high-profile shot at OpenAI, positioning Claude as an "ad-free" alternative [00:05:26 - 00:09:47]. Their campaign contrasted starkly with OpenAI’s announcement that ChatGPT will introduce advertisements across its free and entry-level paid tiers.
Both firms face business model questions:
Anthropic’s Position: A bet on staying ad-free may be difficult to sustain as compute costs drop and open-source competition intensifies.
OpenAI’s Monetization: While OpenAI once publicly shunned ads, economic pressures have shifted strategies.
For business leaders, understanding the long-term direction of AI monetization models helps evaluate vendor lock-in risk and future total cost of ownership.
SpaceX and xAI Merger: Satellite Data Centers for AI Compute
Elon Musk’s decision to combine SpaceX and xAI was not widely anticipated, but creates a new entity valued at $1.2 trillion—a clear play to position for an upcoming IPO and challenge OpenAI and Anthropic’s public market entries [00:15:27 - 00:17:35]. The merger strategy:
Resource Cross-Leverage: SpaceX gains AI capabilities; xAI receives capital and operational scale.
Orbital Data Center Ambition: Plans to launch up to a million satellites as space-based data centers to drive down AI compute costs—a technically possible but long-horizon vision with regulatory and engineering obstacles.
Practically, the immediate impact is financial and narrative: SpaceX seeks to maximize IPO value by embracing AI as core to its future business.
Perplexity Model Council: Multi-Model Answers for High-Stakes Workflow
Perplexity AI’s Model Council aggregates responses from multiple leading models (Claude Opus 4.6, GPT-5.2, Gemini 3.0) and synthesizes them into a single answer [00:21:27 - 00:24:11]. This approach is designed for business-critical tasks like investment research, compliance, and fact-checking, where cross-verifying model outputs reduces the risk of hallucination.
Currently, access is limited to Perplexity’s premium tiers, but this "multi-model" strategy addresses growing enterprise skepticism about relying exclusively on a single AI provider’s outputs.
Coding AI: OpenAI GPT-5.3 Codex and Anthropic Opus 4.6
Two major coding models were unexpectedly launched within one hour of each other, spotlighting the competition in agentic workflows and code generation [00:26:26 - 00:37:07].
Anthropic Opus 4.6: Unlocks agent teams for parallel task coordination, and extends context window up to 1 million tokens—enabling sophisticated use cases in industries with large document sets. Early integration into Microsoft PowerPoint previews the business productivity value.
OpenAI GPT-5.3 Codex: Debuts only in the new native Mac "Codex" app, not regular ChatGPT. Achieves industry-leading scores on real-world coding benchmarks. OpenAI signals a potential dual focus: some advanced models may only reach end users via specialized apps, shaping how organizations plan developer workflow upgrades.
These releases mark a new phase: best-in-class AI capabilities for coding and automation are rapidly evolving, forcing technology leaders to monitor not just public web product updates, but also app-specific and API-first releases.
Competitive Landscape and What to Watch
Meta, after ten months of silence, is on the verge of launching new agents, deeper external integrations, and model upgrades—potentially including direct support for viral open-source agent platforms like OpenClaw [00:17:58 - 00:21:27]. Meanwhile, Google is preparing more aggressive spending on AI R&D, as the recently released Claude and Codex models leapfrog Gemini on third-party benchmarks.
The market is transitioning to a new era of:
Agentic AI for workflow automation
Multi-model cross-verification for critical tasks
Enterprise governance for large-scale deployment
Open competition between proprietary and open-source approaches
Business owners and enterprise decision-makers should prepare for accelerated update cycles, shifting business models, and profound impacts on IT strategy, professional services procurement, and even organizational structure.
Conclusion: The past week’s tightly packed AI news cycle demonstrates that what happens at the platform and model level for OpenAI, Anthropic, and other labs will have tangible implications for every area of business operations, from software procurement to compliance to workforce planning and product differentiation. Staying current on these developments is no longer optional for leaders shaping the future of their organizations.
Topics Covered in This Episode:
- OpenAI Frontier Platform for Enterprise AI Agents
- Anthropic Super Bowl Ad vs OpenAI Ads Debate
- Anthropic Claude CoWork Industry-Specific Plugins Launched
- Legal and Finance Industry Shaken by AI Plugins
- SpaceX and XAI Billion-Dollar Merger Announced
- Meta AI Platform Upgrade and Agent Integrations
- Perplexity Model Council Multi-AI Response Launched
- Anthropic Opus 4.6 Multi-Agent Teams Release
- OpenAI GPT-5.3 Codex Benchmarking Leadership
- OpenAI Codex Mac App Dedicated Launch
- Software Market and SaaS Disruption from AI
- Anthropic Claude Integration in Microsoft PowerPoint
- Google Gemini 3 General Availability Imminent
- XAI Grok Video Model Leaderboard Update
Episode Transcript
Jordan Wilson [00:00:17]:
If I had to choose one word that described what happened in AI this week, has to be catty. I mean, yes, all the big AI labs are constantly competing against each other, but at least anthropic and open AI this week took their, I don't know, pseudo rivalry to a whole another level between Anthropic's Super Bowl ad and OpenAI's Clapback, plus their back to back model releases made for an interesting week between two of the heavyweights in AI. But that wasn't all because, well, SpaceX and XAI, Elon Musk's company merged. Meta may be cooking up something really impactful after essentially a year of silence, and that's just the tip of the iceberg in terms of what's new in AI this week. And if you couldn't keep up, don't worry. That's my job to keep you up to date in the AI news that matters. Alright. Let's get going and go over everything.
Jordan Wilson [00:01:27]:
What's going on y'all? If you're new here, my name is Jordan Wilson, and welcome to Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me keep up with the nonstop AI developments, how to make sense of them, cut through the BS and the marketing, take what matters to grow our companies and our careers. So if that's what you're trying to do, it starts here with the unedited, unscripted livestream podcast. But to take it to the next level, make sure you go to our website at youreverydayai.com. Each and every day, we recap the livestream podcast for today as well as giving you all of the other new AI developments that you need to know. But on Mondays, we go over the AI news that matters. So if you can only tune in one day a week and you really only care about what's new, well, Mondays are the day for you. So, with that, and hey.
Jordan Wilson [00:02:19]:
Live stream audience. Gotta shout shout you guys out more. Great great, great to see everyone joining in. Douglas joining from Indie. Jacob, tuning in here from Austin, Texas. Dennis, good to see you from Raleigh, North Carolina. Oh, my Tar Heels put a I wouldn't say a beating on the, on the Duke Blue Devils, but they at least won. Jay, good to see you from, West Virginia.
Jordan Wilson [00:02:44]:
Joe, everyone else, thanks for tuning in. Alright. But let's start at the top. So this story didn't get a lot of headlines, maybe, but it could be one of the most impactful launches from OpenAI, I don't know, in a while. That's because OpenAI has introduced Frontier, a new platform designed for enterprise customers to build, deploy, and manage AI agents from both OpenAI and third party companies. So Frontier aims to tackle what OpenAI calls agent sprawl by unifying fragmented tools, disconnected workflows, and siloed data. So in Frontier, each AI agent managed there receives a unique identity along with permissions and guardrails, helping enterprises maintain security and compliance in regulated industries. So OpenAI said that their new frontier program is modeled after human workplace principles, such as shared context, onboarding, hands on learning, and clear boundaries to make AI agents more effective collaborators.
Jordan Wilson [00:03:50]:
So the new platform supports agents operating across local, cloud, and open AI hosted environments using open standards to allow integration of OpenAI proprietary and also third party agents all in one system. So OpenAI said that early adopters of Frontier include major companies like Intuit, HP, Oracle, and Uber who provided guidance during the platform's development. OpenAI is also partnering with organizations such as Abridge, Clay, Ambiance, Decagon, Harvey, and Sierra to further refine Frontier base on real customer needs. So, yeah, we'll get to all the the new model releases and all the the back and forth. But, I mean, when it comes to AI in the enterprise, maybe one of the most impactful news stories of the week was actually Frontier platform from OpenAI. So here's here's the reality. I think that the majority of us listening out there may not get access because this is very limited right now just to some of the largest enterprise companies, in the world. But I do think whatever OpenAI kind of cooks up in the frontier, platform will eventually be rolled out, across their other offerings, everything from, you know, chat GPT and and codex, to their agent builder.
Jordan Wilson [00:05:09]:
Right? So I think what happens in the frontier platform with some of their biggest enterprise customers, will eventually impact, you know, hundreds of millions of open AI users. Alright. Our next piece of AI news. Yeah. Let's get into, some of the spicy stuff right away. So anthropic in their Super Bowl commercial and OpenAI clap backed at least online. So Anthropic's first ever Super Bowl commercial has ignited a public debate about advertising in AI chatbots and even drew a direct response from OpenAI CEO Sam Altman. So Anthropic's ad campaign that debuted well, technically, they released it online last week, but it just debuted during last night's Super Bowl.
Jordan Wilson [00:06:02]:
The new ad for the Claudette chatbot promised to remain ad free contrasting with OpenAI's recent announcement that chat gbt will begin showing ads. So, essentially, if you didn't see the ad, I don't think I could play it due to copyright reasons, but they, you know, they were funny ads. So Infropic kind of depicted, different scenarios. Right? They had one with, you know, someone who was quote, unquote using an AI, to, you know, be better in the gym, you know, therapy, and then, you know, at least in Anthropic's, kind of vision of their, of how other AI ads or sorry, how other AI companies are implementing ads. They kind of, you know, had these, you know, the trainer and the therapist, you know, start pushing unrelated ads in the middle of the simulated conversation, during their ad, right, which is obviously not how OpenAI's ads are going to work. So the, the the ads from Anthropic were not exactly truthful on how ads will be rolling out in the OpenAI platform, which is why OpenAI Sam Altman OpenAI's Sam Altman did respond and clapped back on Twitter last week calling the ads dishonest and saying OpenAI would never run ads as depicted in Anthropic's ad. So Altman emphasized that OpenAI's commitment to making AI accessible to billions, not just wealthy subscribers taking a shot at Anthropic while noting that both companies rely on paid subscriptions. So, this is interesting because Infropic also declined to comment directly to the AdWeek article, but explained in a blog post that its revenue comes from enterprise contracts and subscriptions, which is, you know, not ads.
Jordan Wilson [00:07:56]:
So, right now, this I mean, this was a lot of, I'd say, noise, but it was really interesting. And Anthropic pretty much put their future on the line saying that they're not going to make, Claude ad supported. But I don't see, if I'm being honest, I don't see in the long run how they can hold to that promise. Right? Similarly, right, you have to call a spade a spade here. You know, it was about a year, a year and a half ago where OpenAI CEO Sam Altman, he didn't say that OpenAI would absolutely never have ads. He essentially just said, yeah. Ads would be an absolute, you know, kind of worst case scenario or a last case scenario and just kind of voiced his distaste for ads. Yet, here we are.
Jordan Wilson [00:08:43]:
OpenAI did a couple of weeks ago announce that ads are coming to both their free tier as well as their more affordable $8 a month paid tier called, ChatGPT Go. But when it comes to the long term, I don't see how in traffic after making a big splash with this Super Bowl ad, right, kind of, knocking on their competitors for inserting ads into certain conversations. I don't see how anthropic can stick to this over the long term. Let's be honest. Right? One of their biggest advantages right now, anthropic, well, is their coding tools, and their API, which both of those things, I think, are going to get disrupted by OpenAI. Some more on the cat fight later and how that, aspect is going to specifically play into, Anthropic's short and medium term plans, but also just the price of compute. Right? Inference cost, everything is going to be going down aggressively. I think some of the open source models, have closed the gap.
Jordan Wilson [00:09:47]:
Right? We talk about, you know, OpenClaw. Right? The new, very viral, kind of, you know, AI bot platform. You know, recently, they've been saying that some of the open source models are being used more, than Claude's models, which is kind of what the OpenClaw platform was built around. It was built kind of around Claude's models, and now they're not even the most used models because over the past two weeks, some of the Chinese open source, competitors are closing the gap, right, between open a, or sorry, between, you know, anthropics models and what they can, you know, achieve on the back end via the API for coding agentic tasks, etcetera. So I don't know. Pretty pretty big, pretty big bite, that Anthropic took here taking out a Super Bowl ad to say, yeah, we're not going to do ads like competitors. I don't know how that's gonna hold up. Alright.
Jordan Wilson [00:10:41]:
More new stuff from Anthropic, and this one is pretty big. And again, I think a, news story that kind of slipped under the radar. So Anthropic has released a new set of plugins, and it's quietly shaking certain industries such as legal and finance. And it's impacted a lot of these software stocks. So shares of major software and data services companies plunged late in the week after Anthropic launched new industry specific AI plug ins for its Claude CoWork platform. So Claude CoWork's new plug ins offer specialized tools for sales, finance, data marketing, and legal sectors, letting users automate tasks like file management, document drafting, and folder organization. So the announcement about some of these new platforms, I think the two that really shook, the market were the, the finance and the legal plug ins, and it triggered a almost 6% drop in the main software industry's ETF on Tuesday, its single worst day decline since April. Also, Thompson, Reuters stock suffered its biggest one day loss ever, and, I mean, it's the writings on the wall.
Jordan Wilson [00:12:07]:
This was because of the anthropic release as it fell, more than 15% on Tuesday, while legalzoom.com tumbled almost 20%. So investors are worried that companies may now rely on these AI powered plug ins instead of paying for multiple expensive external software tools directly threatening the software as a service business model. So Wall Street's reaction reflects fears that AI could replace not just the software products, but entire workflows and, obviously, the human hours behind those workflows, putting pressure on both legacy software providers and some of the, very pricey industries that use some of this software and some of what it automates. Right? Specifically, I think we saw, some pretty, I don't know, doom and gloom type news, from the legal industry and the finance industry. And analysts are cautioning though that the sell off is mainly driven by uncertainty about AI's real world impact, suggesting the market may recover as the long term effects become more clear. So, last few audience, what do you what do you think about this one? I don't know if people have been following, this, but right. Going back to, I don't know, about thirteen months ago when we came out with our, 2025 AI prediction and roadmap series, I said this. So maybe I was a little early.
Jordan Wilson [00:13:44]:
Right. But I did say that AI's advancements are going to send shock waves, through specifically, I talked about the legal and finance industry. So, maybe we didn't get that in 2025, but here we are, you know, six weeks into 2026, and it is coming to fruition. And I think this is a story to continue to watch, not just on the legal, and finance industries, but also on the consulting side. I do think that, professional services, especially high priced, professional services are going to go through a complete makeover here in 2026. And in Anthropics, kind of their plug ins, even though this was maybe a bullet point on the radar last week. I mean, we're already seeing some pretty significant, you know, shake ups both in the software industry and in the financial markets. You know, comment here from, YouTube.
Jordan Wilson [00:14:43]:
Lisa saying lawyers are cautious due to our duty of confidentiality, so probably won't jump on this, right away. So, yeah. We'll see. I think some of the early adopters, of this are going to take some risk, but I also think this is going to lead, to a new class of AI native companies that are doing this work, that are doing finance, legal, and just, you know, some of those high priced professional services work. So I think if, essentially, either the big players are going to have to, change their business model or they're going to start to get gobbled up, and they're gonna have to, you know, lay offs in mass. Right? I'll say the quiet part out loud. Alright. More on the markets.
Jordan Wilson [00:15:27]:
Well, right now, the, markets will soon tell us what they think of the new SpaceX and XAI merger. That's because Elon Musk has combined two of his companies in SpaceX and his AI company XAI, forming one of the world's largest and now most valuable private companies, which is privately valued at $1,200,000,000,000 according to Bloomberg. So, this kind of new joint effort comes as SpaceX prepares for a highly anticipated IPO later this year, positioning itself to compete with expected public offerings from AI rivals, OpenAI, and Anthropic. So the merger gives SpaceX access to x AI's advanced AI capabilities while x AI benefits from SpaceX's resources and a significant cash cash infusion. Musk announced ambitious plans with the merger to launch a constellation of a million satellites to create orbital AI data centers, aiming to make space the lowest cost way to generate AI compute within two to three years. Industry experts say the idea of space based data centers is technically possible, although not really feasible with today's technology, but will likely take many years to become practical, probably more than the two to three years that Elon Musk was saying with significant engineering and regulatory hurdles ahead. So some analysts suggest this new merger is also a strategic narrative move from Elon Musk to boost SpaceX's valuation ahead of its IPO, framing it as a leader in both space and AI innovation. So the combined company now includes AI, rockets, space based Internet, and direct to mobile communications in the social media platform, x, which xAI acquired last March.
Jordan Wilson [00:17:35]:
So, yeah, this is just, more of Elon Musk's company changing names, getting new acquisitions. I think most people saw something like this coming that eventually Elon Musk was going to bundle all of his different companies, under one umbrella, to go to come up with a big, IPO in 2026. Alright. Here's one that has been quiet. Right? Elon Musk has obviously not been quiet, with any of his companies, but Meta has been surprisingly quiet for the last ten months because aside from a $14,000,000,000, kind of acquihire of Scale AI and its CEO in a complete shakeup of their own internal AI teams, Meta has been completely silent on the AI release front. Right? So after their, Meta llama four release of spring twenty twenty five, we haven't really seen anything new, in terms of LLMs from Meta in ten plus months, but that may change here in a couple of weeks. So according to leaks from testing catalog, Meta AI is preparing a sweeping platform upgrade, focusing on new agents, deeper integrations, and a new large language model. Also, the acquisition of Manus AI, which Meta completed last month, is leading to the development of a Manus agent and a dedicated browser agent reportedly called Sierra.
Jordan Wilson [00:19:14]:
So these agents are designed to handle complex tasks and automate web browsing, expanding what users can do with meta AI. Also, according to leaks, this new Meta platform could have an OpenClaw integration. Yeah. So the, you know, Clawbot turned Moltbot turned OpenClaw. Right? The, viral, kind of open source agentic AI platform that's been taking over the Internet over the last few weeks reportedly could be integrated directly into Meta's new platform by allowing users to connect any AI model with their own API key. Also, there's the new reported a, Avocado AI video model is reportedly being readied as Meta's primary AI engine appearing in both standard and thinking modes. So Meta is also revamping its website and apps, releasing a new interface with effort selectors for fast and syncing models, offering users a choice between quick replies or deeper reasoning. Alright.
Jordan Wilson [00:20:26]:
There's more that's reportedly happening, under the hood on Meta's new, platform. So there's a new widget prompt, an sorry. A new widget that prompts users to link services such as Gmail, Google Calendar, Outlook, and Outlook Calendar. So this signals that Meta is using a, MCP connector that could go live soon. There's also internal code references according to testing catalog that show Meta is benchmarking other top models like Gemini, Chat GPT, and Claude possibly for multi, multi model routing or performance comparison. So all of these updates are expected to roll out soon, possibly as early as this quarter. Timing that could determine Meta's competitive position as major rivals are also launching new models. Alright.
Jordan Wilson [00:21:27]:
Speaking of models, Perplexity finally released something that I also predicted in 2025, maybe a couple months too early. So Perplexity has just launched their new model council to combine multiple AI models into one verified answer. Yeah. So something I predicted last year called the mixture of models. We're finally seeing it now from one of the big players in AI, much different than a mixture of experts, kind of set up that a lot of the new AI companies use. So according to perplexity, their new model council is a feature that runs a single query across multiple frontier AI models at once and synthesizes the results into one response addressing growing concerns about model bias and uneven performance. So the model council automatically compares outputs from models such as Claude Opus 4.6, GPT five two, and Gemini three point o, then highlights where the use, sorry, where the models agree or disagree, removing the need for users to manually cross check performance. Well, at least I wouldn't, completely take that away from the users, but at least I think helps in that process.
Jordan Wilson [00:22:51]:
So perplexity says its internal data shows that the AI model performance varies widely by task with some models excelling at coding while others perform better at research or creative work, making this multi model verification increasingly important. So the feature is positioned for high stake use cases like investment research, complex career or business decisions, and fact verification where a single model's blind spots, could lead to costly mistakes. Or, if a single model hallucinates, you hope that the other models in this multi model setup could spot that. So right now, unfortunately, the model console is only available to perplexity max subscribers, so on their higher tier plan on the web with mobile app support coming soon. That wasn't the only, perplexity update this week. They also launched their upgraded deep research tool. So their new deep research, capability positions it as a faster and more accurate way to handle complex real world research task. Their new deep research is also only available to pro users, and they, announced kind of a new, deep research benchmark called, Draco.
Jordan Wilson [00:24:12]:
Obviously, it's their own internal benchmark across the deep researches, so their model did perform best. So we'll see. I don't know if that picks up any steam. You know, hey. Live stream audience, podcast audience as well. I always go back and, you know, check comments on Spotify. Should we be diving more into, you know, perplexity deep research or, you know, some of these new perplexity, features. I don't know.
Jordan Wilson [00:24:36]:
I think that anthropic, has shifted a lot more over the past three months to non coding, tasks, which is, at least for me personally and probably for a lot of our audience, maybe more impactful, and they are finally rolling out. Anthropic is to kind of base users where, you know, perplexity, a lot of these seem really impressive, and probably worth spending time on, but, you know, you do have to be on the higher, paid tier right now, for a lot of these new features, including the mod of console and their new upgraded deep research. So I just don't know how much of an appetite there is. Like, how like, what percentage of people, you know, have the perplexity max plan versus now the, you know, anthropic $20 a month plan. So, yeah. Joe here says, I think perplexity is, way too far behind. I'll say yes and no on that, Joe. So I think with some things, they absolutely are.
Jordan Wilson [00:25:35]:
I think that they're behind on bringing their best technology to the majority of their paid users. But, I will say I have been impressed, more recently, with Perplexity's offerings. I do think that they are pushing the frontier a little bit, but, I think this is also kind of what plagues Anthropic in early twenty twenty five. Anthropic had some amazing updates that they only rolled out to their Macs or their enterprise users. So kind of the same thing, you know, kind of the go to market, strategy for here, some of these bigger companies. You know, I don't know. You don't really get a second, shot at some of these announcements. So, you know, when they do roll it out, to kind of, quote, unquote, normal paid users, I think maybe a lot of the hype or the interest will have died down at that point.
Jordan Wilson [00:26:26]:
Alright. So we started this week's recap off by talking about some of the cattiness between anthropic, and OpenAI, at least when it came to, kind of the Super Bowl ads and OpenAI's response. But it actually played out again because both companies released new models this week and, coincidentally, within an hour of each other. Yeah. I don't think that was a coincidence. So, let's first we'll go through it chronologically, I guess, because Anthropic was first with their new Opus 4.6 and then OpenAI responded. But let's first get to what's new in Infropic's new Opus 4.6 model, which I've been using a ton and in very impressed by. So Infropic has released Opus 4.6, its most advanced AI model to date, marking a significant upgrade from Opus 4.5, which just was debuted in November.
Jordan Wilson [00:27:26]:
So a faster, release cycle for Anthropic as well in choosing this time, to come out with the 4.6 variation in their Opus model first. So, again, yeah, not to get too technical, but, Opus kind of has or sorry. Anthropic has three tiers. Their haiku, which is their, kind of least intelligent but fastest tier, their Sonnet, which is kind of the middle, and then their Opus, which is their most powerful. So, they led with 4.5 with Sonnet, and then later released Opus 4.5. So a lot of people were assuming we would get a Sonnet four point six first, but we actually got a Opus four point six. So they're kind of, switching up their, release cadence or their priority here. But the new standout feature from Opus 4.6 is definitely agent teams, which lets multiple AI agents split and coordinates complex tasks working in parallel to finish jobs faster and more efficiently.
Jordan Wilson [00:28:27]:
So the new agent teams feature is currently available as a research preview for API users and subscribers. So as an example, if you're using, Claude Co work or Claude Code on the Mac app like I am, you'll see this new agents teams feature, which is actually really fun to watch for me. Right? I was using, Claude code with Opus, 4.6 a little bit over, the weekend and just kind of watching kind of these kind of sub agents, not, like, argue with each other. Right, but kind of the made the main agent was checking some of the sub agents work and was like, no. This is wrong. Let me go send a third agent to go, verify this. So it is really interesting to kinda see this multi agent orchestration, without the user having to do anything. So this is pretty a, a pretty big step change, I think, in how multi agent orchestration works by default and kind of de facto, now under the hood.
Jordan Wilson [00:29:27]:
The other big, highlight from Opus 4.6 is that it now supports a context window of a million tokens, on the API side, allowing it to recall and process much larger volumes of information. I will say this, at least, testing this in Claude code, kind of side by side with codex, in their, OpenAI's new GPT five three codex. I don't know. I wasn't seeing this new expanded memory. I was actually seeing much better, memory within the same, sorry, context window, in GPT five three Opus or sorry, GPT five three codex, much better in terms of memory and context window in my personal, testing than Opus 4.6 was. So I think, again, this is probably more if you're using it on the API side. But, there's more news with Opus 4.6. That's because it is already directly integrated into some Microsoft products.
Jordan Wilson [00:30:27]:
So the model is integrated directly into Microsoft PowerPoint as a side panel, letting users create and edit presentations with Claude's help inside of PowerPoint instead of jumping between apps. So, that may be one of the it's kinda like a like a post note, right, on the Opus 4.6, release. So, yes, this is available on the front end. If you're using claw.a you know, claw.ai, as a front end user, you will now see Opus 4.6. You're not really gonna really see that, agent team framework on the front end, but you will see it if you're using, the, as an example, Claude code or Claude cowork. You will be able to take advantage of those Opus, you know, 4.6 agent teams. But like I said, one of the more impressive things or one of the most immediately useful things might be the integration into Microsoft PowerPoint. So Anthropic says this direct integration streamlines workflows for knowledge workers, making AI assistants more accessible and practical for everyday business tasks.
Jordan Wilson [00:31:37]:
So let's talk about the timing. Okay? So it was less than an hour after Anthropic released Opus 4.5 that we got our next and last big AI news story of the week. So was this an intentional clap back? I I don't know. Right? Was someone at OpenAI, right, there there are always two or three releases ahead. So, you know, I think at any, time in point, any of the AI labs can, you know, release their next, you know, announcement. Right? GBD53 or Opus four seven. So it was pretty interesting after Anthropic just went straight for the jugular with their Super Bowl ad, essentially attacking OpenAI for having ads in their chatbot that OpenAI responded by I don't know. Let me be honest.
Jordan Wilson [00:32:31]:
Wiping the floor with Opus 4.6. Right? Opus 4.6 comes out, yes, impressive benchmarks. But out of nowhere, OpenAI comes with GPT five three codex, which from the benchmarking and coding side, OpenAI means business now. And even OpenAI's, Super Bowl ad was codex. It wasn't anything about chat GPT. So it looks like OpenAI is making, I wouldn't say a pivot, but they are making coding, and agentic benchmarks a huge priority. So, let's talk about the new model from OpenAI GPT five three codex. So, OpenAI just released that GPT five three codex, their most advanced coding agent to date.
Jordan Wilson [00:33:24]:
So according to OpenAI, GPT five three codex seeks new industry records on industry benchmarks, scoring a 77.3% on terminal bench two point o, a 13 leap over its predecessor, GPT five two codex, and about 10 points ahead of Opus 4.6. That's the thing that was absolutely bonkers to me. Right? Opus, 4.6 comes out, obviously, for a couple of minutes. It is the industry leader on one of the most important, benchmarks, terminal bench two point o. And then g p t five three codex comes out of nowhere, and just slaps Opus four, six and the whole industry silly, on the coding side. So the new codex model though is not just for coding. It can handle a wide range of computer based tasks from debugging and deploying software to building presentations and analyzing spreadsheets, aiming to automate much of the software development process. So, OpenAI describes GPT five three codex as its first model to play a key role in developing itself.
Jordan Wilson [00:34:42]:
That's a big, a big, thing to highlight as well. Right? We've talked about in previous episodes, that Anthropic's, very successful clogged code now essentially just updates itself. So OpenAI did, say that this was the first time, that one of their models essentially helped develop or led a, played a key role in developing itself. So, this came out kind of hand in hand with OpenAI's new codecs Mac app. Alright. So if you wanna use this new model, you're not gonna find it going to chat gpt,uh,.com. You can use its codecs online, but the other big release from OpenAI this week, well, they officially launched a dedicated codex Mac app, this week as well. So this is designed as their new command center for agents, a native desktop application that allows developers to manage multiple AI coding agents simultaneously and automates complex workflows.
Jordan Wilson [00:35:51]:
So, this was an interesting move from OpenAI because I don't think we've seen this before, that they've kind of come up with a new model. Right? So in this case, going technically from GPT five two to GPT five three, but only the codex variation. Right? So normally, whether they're going from g p t five to five one, from five one to 5.2, you usually see that debut in their main app, right, inside chatgpt.com. But like I said, you're not gonna find gpt5three, in chatgpt.com. You have to use it in codex. And then like I said, OpenAI specifically promotes and highlights codex during its Super Bowl ad. So interesting here that we're almost getting now maybe a dual focus, from OpenAI when it comes to bringing different models, to market, which I don't know if I'm a huge fan of that. Right? Because like I said, I think that there's probably millions of people out there that could benefit from using this new model, GPT 5.3 codex, for non coding reasons.
Jordan Wilson [00:37:07]:
Right? Creating presentations and spreadsheets, and it's really, really good at those things. But I think for the most part, people look at codecs even by name only, and they're gonna say, oh, I'm not a software engineer. I'm not a coder. So, you know, I'm not gonna use codecs. I'm just gonna wait for something to, you know, release inside of chat g p t. So who knows if we're gonna get a, a version of a g p t 5.3 anytime soon inside of chat g p t. But if you wanna use it for now, you do have to, use it via codex either on the web or on their newly released platform. Alright.
Jordan Wilson [00:37:44]:
That's not all. Yeah. There was a lot, that happened this week. A lot of big AI news stories, but that's not it. So now we're going to, end the show with our quick bullet points recap, going, you know, some of the smaller stories, but also some some rumors, some leaks, etcetera. So, here's what you need to know. So, Alexa Plus became generally available to everyone in The US. I've been using it for a couple months.
Jordan Wilson [00:38:16]:
Not a fan. Also, cloud, clawed in PowerPoints available in research preview, for Max team and enterprise subscribers like I talked about. A new $200,000,000 Snowflake and OpenAI deal integrates g p t 5.2 into Snowflake for enterprise AI agents. Google, reportedly plans to double capital spending on AI next year. OpenAI hired former anthropic researcher Dylan Scandonaro as their new head of preparedness. In an interview, Sam Altman suggested that AI could eventually replace him as OpenAI CEO. Yeah. That one was weird and interesting.
Jordan Wilson [00:39:02]:
Anthropic, also had an update to Opus 4.6 with Opus 4.6 fast mode available as a preview available in cloud code. But my gosh, that thing yes. It's fast, but you don't get a lot of usages on it. Very strict rate limits on it right now. Google introduced AI expanded access add on, with higher workspace AI usage limits. XAI released a new version of their Grok imagine video, which did take top on the image to video leaderboards. Anthropic partnered with the Allen Institute deploying Claude agents to automate lab tasks. The AI bot only social network, mult bot multbook reportedly exposed over 1,000,000 credentials according to a, security research firm, arena, formerly called LM arena.
Jordan Wilson [00:39:58]:
They came out with a new max mode that intelligently routes prompts to the best AI model using 5,000,000 plus community votes to decide where your prompt should go to. So it looks like LM arena may be turning into, kind of like a perplexity type platform that you can actually use and not just use to test models. Speaking of testing, Google is reportedly testing personal intelligence in notebook LM, one of my favorite products in most use. OpenAI has their codex model working on Windows internally, though there's no release date yet. So, the new codex app was released Mac only, but we, we'll be getting a Windows version soon. Perplexity has improved their memory feature for Pro and Max users. Google is reportedly preparing their Gemini three, for, general availability. So, yeah, this is gonna be, something to keep an eye on as, the new Opus four six and GPD 5.3 codex have kind of stolen Gemini three's spot on a lot of the different leaderboards.
Jordan Wilson [00:41:09]:
We'll see how Google responds with their Gemini three GA version. OpenAI is officially retiring the standard selection of the GPT four o model this week. Goodfire raised a $150,000,000 series b, and Apple is planning to release a CarPlay update that will allow you to use third party AI chatbots in the car while retaining Siri control. Alright. That was a lot happening in the world of AI. So I hope that today's recap was helpful. Like I said, we do the AI news that matters every single Monday to keep you up to date, but we do this thing every single day, Monday through Friday. So, if this was helpful, please take ten seconds.
Jordan Wilson [00:42:00]:
If you're listening, here on the live stream on LinkedIn, to click that repost button and repost this to your network, I would really appreciate this. That's how we keep this, free and unbiased and try to make you the smartest person in AI at your company. It comes from you sharing this with others. If you're listening on the podcast, hey. Take ten seconds. Make sure you're following and subscribe to the show whether you're listening on Spotify or Apple Podcasts, and please leave us a rating. That really helps us help more people like you to make sense of everything that's happening in the world of AI. And then when you're done doing those things, please go to your everydayai.com.
Jordan Wilson [00:42:36]:
Sign up for the free daily newsletter. Thank you for tuning in. Hope to see you back tomorrow and everyday for more everyday AI. Thanks, y'all.