← Everyday AI Podcast – An AI and ChatGPT Podcast
Everyday AI Podcast – An AI and ChatGPT Podcast
ChatGPT’s Image 1.5 winning, Google launches Gemini 3 Flash, Meta going after Google and more
00:34:42
Loading summary
Transcript3 lines
- [00:01]
A
This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.
- [00:16]
B
It's the end of the year, so maybe things have started to wind down for you or your company and that's understandable. That's usually what happens, especially here in the US with the holidays around the corner. But don't sleep on what happened in the AI world this week because while maybe many of us are starting to wind down, if anything, the AI wars have only started to heat up. I mean, just this week we have a new multi billion dollar partnership, a new large language model update that may impact billions billions of Internet users, and a new image model that at least by the rankings is the best in the world. All right, if you're not ready to clock out yet for 2025 and you want to know what's happening in the world of AI without wasting hours every single day, we got you. Let's get into it. Welcome. My name is Jordan Wilson and this is Everyday AI. If you're new here, this is a daily live stream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with what's happening in the world of AI, but how we can make sense of the non stop updates because they are non stop and just extract the important insights to grow our companies and our careers. It starts here with the unedited, unscripted live stream podcast. But to really take it to the next level, your cheat code is our website, your everyday AI.com. there you can obviously go watch and listen to more than 670 episodes, all sorted by category. But you're going to go want to sign up for our free daily newsletter where each day we recap the podcast and we give you the AI news that you need for the day. All right, so FYI, if you missed it last week, make sure you go back and listen to episodes 674 and 676 are 2025 AI roadmap rewind. It's extremely important they're quicker episodes so make sure you go back and give those a listen. Also couple housekeeping things. So the newsletter and podcast will be taking the short break at the end of this week from Wednesday to Friday. Yeah, with Chr, Christmas Eve and Christmas and like I said, most people especially in the US winding down, we at Everyday AI are going to as well. So we're going to be also for the rest of the year playing some of our top podcasts from the year, so don't miss that. So maybe you missed something back in June or an important interview back in May. We're going to be playing some of our favorites as we cook up something special for you, so make sure you stay tuned for that. All right, let's get into the AI news that matters now for the week of December 22nd. All right, first, a big one here, a $10 billion partnership between two of the biggest names in tech that not a lot of people are talking about. So According to reports, OpenAI is in confidential discussions with Amazon about a potential investment and an agreement for OpenAI to use AWS's AI chips. So according to reporting from the information, the talks could include an investment that might exceed $10 billion from AM to OpenAI, though details remain fluid and are subject to change. So the information first reported the details and then some more information here from the CNBC report that I'm showing on screen. So the discussions follow OpenAI's October restructure that gave it greater flexibility to raise outside capital and to enter partnerships for compute beyond its primary backer, Microsoft. So Microsoft has invested more than 13 billion in OpenAI since 2019 and maintains a major, major partnership, but it no longer has exclusive rights to be OpenAI's compute provider after those October changes. So Amazon already has large investments in the AI startup scene, aside from their own Nova models and obviously their AWS presence, including at least $8 billion invested in Anthropic, and is also expanding its AI chip strategy with its new Trainium line. So OpenAI has also signed these similar deals. And you know, there's a lot of scrutiny out there, right, when people are like, oh, OpenAI raised $10 billion. But sometimes you got to read the footnotes and it's like, oh, well, it's mainly just to, you know, buy these chips or to use, you know, these certain chips. So, you know, they have a kind of similar deal with other big providers such as Nvidia, AMD, and even Broadcom, where they're reportedly working on their own in house AI chips. OpenAI recently completed a $6.6 billion secondary share that valued the company at about $500 billion. But there's new reporting this week that OpenAI is out to raise another fresh set of funds that would value the company at more than $800 billion. So pretty big news here. And OpenAI even came out with a lot of people found it kind of strange, right? They put a video out that said, hey, we've had to pause research. We have all these, you know, Viral, you know, products, right? And they're like, we don't have enough compute. So maybe this is this new $10 billion partnership with Amazon is something that will help them scale in 2026 and beyond, as Google has really started to pick up the steam and catch OpenAI on a lot of fronts. All right, speaking of catching Google, Meta is reportedly coming out with their own version of Nano Banano. Nano Banana. Sorry. So according to reports in the Wall Street Journal, Meta is preparing to release Mango, a new image and video model, and then Avocado, which we mentioned last week, their new generation text model. So meta's Super Intelligence Labs, which is led by former Scale AI founder Alexander Wang, is reportedly developing Mango for images and video and Avocado for text, with both models expected to be released in the first half, 2026, according to the Wall Street Journal. So the plan launches come as Metis tries to close the gap with rivals in creative AI, especially Google, OpenAI and Adobe, which already offers strong image and video generation tools. And I mean, if you've been paying attention at all, I mean, Google's recent Nano Banana image generation image generation tool has just gone absolutely viral and has been one key factor at least in Google really closing the gap with OpenAI when it comes overall users. So Meta's move highlights how the consumer AI race has kind of shifted from just text chat to more just trying to create viral visual content or at least useful AI visual content. And we did talk about this last week. These new models from OpenAI may reportedly be proprietary, right? So their previous LLAMA models have been essentially open source models, right. That you can use for free under most conditions. So reporting has said that Meta may be shifting their strategy and they've been kind of essentially quiet for nearly eight months with no major releases, at least on the LLM side. They've had a lot of releases on their SAM side. The segment, anything but it's kind of been quiet on the large language model text front. I don't even think that their Llama 4 behemoth model was actually released. I'll have to go back and check that and let you guys know in the newsletter. But yeah, it's been kind of quiet on the Meta front ever since they've been spending tens of billions of dollars on these Aqua hires, these acquisitions, paying researchers reportedly, you know, hundreds of millions of dollars over short term contracts. So I, I don't know, we'll find out, I guess in early 2026 if met is going to come out cooking or if it's going to Be a swing and miss, but we do know at least where they're trying to compete. All right, next piece of AI news. And this may seem like a small little update, but I think it's actually going to be bigger and might be shifting my viewpoint on Anthropic in the new year. So Anthropic has just widened access and released their Claude Chrome plugin to anyone on a paid plan where previously it was restricted to only those users on the Max tier or on enterprise tier plans. So the cloud. Claude Chrome plugin, yeah, which runs on the cloud. But Claude Chrome plugin lets paid Claude users install the extension and have Claude interact with websites. Which means the model can now fill forms, manage email and calendars and complete multi step workflows on a user's behalf. And this is pretty big, right? Not just, you know, finally rolling out access to all paid users, but even just their approach here because we know that Claude has some of the best computer use and a agentic models in the world with their new 4.5 releases. So a little bit different strategy here when it comes to the agentic side for consumers, right, where you see perplexity and OpenAI taking more of approach of releasing agentic browsers in Comet and Atlas respectively, and Claude on the other hand going the Chrome extension route. So I have been testing it for a couple of days. There's some things I think it does pretty well and it does maybe just as well as Perplexity or Atlas. And then there's other things that I think it maybe falls short on and that would maybe be better suited for an agentic browser. Yet at the same token, right, we've heard OpenAI kind of, you know, had to pause or slightly slow down a lot of their other projects, including reportedly the Atlas browser, although it's back to being updated fairly regularly right now. So a little bit different, you know, strategy here with Anthropic, but I think it's encouraging to see some of their later moves in 2025 maybe signaling a shift that they are not necessarily abandoning the everyday non coding consumer and non coding teams. Right. That's kind of been their specialty is just anything on software engineering, development, coding. Right. So now it's encouraging to see Anthropic rolling out, you know, still yes, little technical updates, but ones that are not necessarily just for people in the software engineering and related spaces. Foreign. Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more but can't really get traction to find ROI on Gen AI hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead of and some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everydayai.com partner to get in contact with our team or you can just click on the partner section of our website will help you stop running in those AI circles and help get your team ahead and build a straight path to ROI on Gen AI. All right, next piece. And I didn't even tease this one in the opener and it might be one of the biggest news AI news stories of the year technically, because this is either going to be a big splash or a flop. We'll see. So OpenAI has officially launched its App Store and is now accepting third party app submissions. So yes, this week they did launch their App Directory, so you can go find it at chatgpt.com apps making it possible for ChatGPT's 800/million users to search, discover and install apps that run directly inside of conversations. So developers and here's the new news aside from the App Store is now live and I think last count there was maybe 35 or so apps available depending on what country you're in. But the big news here is now developers can submit apps, whereas before OpenAI had a very kind of controlled rollout with just some trusted partners. But now app submissions are open to third party developers and those may be approved as soon as early 2026. So if you are a developer you can submit apps through the OpenAI developer platform. You can track the review status and you must meet OpenAI's usage safety and privacy guidelines before any apps appear in the App Director directory. So if you're trying to use apps, well, once you install them, and for some of them, obviously you would need to connect your account. So if you want to use the Canva app as an example and design things inside of ChatGPT, well, you want to connect your Canva app, but once installed, users can just invoke apps by hitting the APT button and then clicking it. Or sometimes, depending on what you say, OpenAI will just assume, oh, you want to use this app, certain app. So there's some early restrictions. Right now, apps may link out to complete purchases, but only for physical goods. And selling digital goods, subscriptions or in app services is not available right now. Also, all apps must be suitable for general audiences, including teens, and include a clear privacy policy and support contact. So, yeah, and if you haven't been paying attention the last couple of weeks, there's actually been some fairly useful apps that have been released. Right. So we covered last week Photoshop released a handful, including their Photoshop app. But here's some other ones that just dropped. I mean, now you have GitHub, Replit, Gmail, Lovable, Mailchimp, Stripe. So, yeah, a lot of other apps that have just launched. All right, next big piece of AI news, and this is the one that will technically impact billions of Internet users, and it's actually much bigger than you may think. So Google has launched Gemini 3 Flash, a faster and cheaper version of its Gemini 3 Pro model. And here's the big piece of news. They made it the default model in the Gemini app, even for free users and in AI mode in search, also for free users. So, yeah, even if you are a free user, if you're using Google Gemini or if you're just using AI mode in search, now it is run by Gemini 3 flash, which is pretty cool because the benchmarks on Gemini 3 flash are very impressive. So Google says Gemini 3 flash is faster and cheaper to run than previous Flash releases, and it will replace eventually Gemini 2.5 flash for routine tasks nationwide. So the new model adds multimodal reasoning. It can watch videos and back images, listen to audio and even read text. Right? Then turn those inputs into content and answers. So Google also, according to their internal benchmarks, which is very impressive. So this is the smaller, faster version of Gemini 3 Pro, which was released just over a month ago. Yet Gemini 3 Flash outperforms Gemini 3 Pro on some key benchmarks, including SWE, Bench Verify, which is a coding agent benchmark. So a lot of people were wondering on Twitter and elsewhere, okay, how is this brand new model, you know, outperforming the bigger version, the Pro version, on one of the most important metrics going into 2026, which is, you know, sweet. Bench Verified, which is a coding agent benchmark. Right? And essentially some Google staffers have kind of said that this model is not a distilled version of the bigger model, which a lot of people assume, oh, the big model, they build that in the Flash or the small or the mini, Right. Depending on which, that you're looking at is just a distilled version. Right. It's a watered down version of the big one. And that's not necessarily true in this case. Right. Because Google did say that they applied some new reinforcement learning techniques to Gemini 3 Flash, which allowed actually some of the benchmarks to improve over the model that they just released a month ago. And they also did denote that Gemini 3 Pro will probably be getting a couple refreshes even before the generally available model. So yeah, you're not going to see a Gemini, probably not going to see a gemini, you know, 3.1 pro or anything like that. But we will see versions of Gemini 3 Pro that are better and more capable than the one that we got in mid November. And like I said, this is big because we've heard that eventually Google may default the Google search to AI mode. Right. So it running on Gemini 3 Flash is literally potentially bringing one of the now world's most capable models. Right. Gemini 3 flash is a top five model on artificial analysis. Yes. The small, cheap, fast version is extremely powerful. And like I said, this could be rolling out right when and if Google does make that transition to billions of users worldwide. All right, more intelligence for everyone. Yum, yum, yum. All right, next piece of AI news. Speaking of intelligence, but this time at the the federal level. So the US Department of Energy has announced agreements with 24 tech organizations to collaborate on their Genesis mission, a government led effort to use advanced AI systems to accelerate scientific discovery, strengthen national security and drive energy innovation. So we did cover the Genesis mission a couple of weeks ago when it was first announced, but now we know which third parties are going to be involved in this domestic project and that includes some of the the biggest names in tech. Right. So Amazon Web services, Google, Microsoft, Nvidia, Intel, IBM, AMD, OpenAI, Anthropic, Oracle, HPE, Dell, Cerebras, Grok, the inference company. Not Xai's chatbot, Core Weave, Palantir. Right. Several others. So essentially any big name in tech or any big name in AI is involved. I don't know if I read off Anthropic and Oracle on that list, but they're in there too. So the announcement follows President Trump's executive order on removing barriers to American leadership in AI and also supports the administration's America's AI Action Plan aimed at reducing reliance on foreign adversaries. So the Department of Energy says the Genesis mission will unite companies, universities, nonprofits and national laboratories to accelerate research productivity and just really push the US on science, innovation and AI development. So the DOE specified that any tool or products developed for Genesis Mission will be architecture agnostic. In other words, they'll be designed to work across different hardware and software systems, which could broaden access for researchers and company. And obviously the amount of data that is going to be available is amounts of data that we have not, not generally seen for frontier models. So this is actually exciting and we'll see how this develops. We don't know a lot about these new partnerships yet, but it's going to be unfolding. And hey, spoiler alert when we go into our 2026 AI roadmap series. Sabrina AI is definitely going to be at the top of the list in this Genesis mission does play into it and I think this does impact everyone, right? Even if you're not an international company, whether you realize it or not, the AI race, it's, it's impacting people's 401k, especially here in the U.S. it's, it's impacting national security, it's impacting energy, right? So it's impacting things at home that you may not even be aware of. So this is something that we're definitely going to be keeping an eye on over the next couple of months. All right, next piece of AI news. Everyone's favorite AI tool, right? I don't know, it's, it's, it's a top toss up for me, but I can't go through more than a day or so without using NotebookLM. And NotebookLM has announced a series of pretty big product updates this week. Kind of snuck them in like every, right when everyone's logging off for 2026. Notebook LLM is like here, let's slap you with a bunch of big updates. So one of the, the biggest one is now it runs on Gemini 3. So previously Notebook LM was powered by Gemini 2.5. Now it is powered by Gemini 3, Google's most capable model, improving reasoning and multimodal understanding compared to prior versions and promising more accurate synthesis of text, images and unstructured data. All right, a couple more big updates, the next one, and this one is maybe just as big. Well, Gemini and NotebookLM integration is now live for many paid users at least. So if you go into your paid account and if you click the plus button in NotebookLM, you'll probably see, or, sorry, in Google Gemini, you'll probably see a new Notebook LM tab. And then you can upload notebooks into the Gemini app, enabling combined notebook projects, generations of images or apps inspired by your notebook Research and continue development of notebooks with integrated web research tools. So, at least for me, this didn't go to any of my paid workspace accounts, but it did go to my paid personal account. So we'll see if that remains true. I'm hoping it does roll out to all paid accounts, but I would check yours as as well. Just right now, go into the Gemini app app, click that little plus button there in Gemini. And if it's there, there will be a new Notebook LM tab. So where you would normally see things like, you know, upload files, add from drive, import code, there will or should be a new button there if you're on a paid personal plan. All right, that's not all. There's two other updates. So one was data. Tables are available now to all Pro and ultra users, with free users getting access in the coming weeks. So tables, essentially, users synthesize facts across sources, find hidden insights, and export tables directly to Google Sheets for further analysis. And this is huge, right? I've been talking so much, right, on our 2025 AI roadmap Rewind, I talked about AI's impact so far on the consulting industry and its upcoming and forthcoming impact. And this is another small but big thing, right? Notebook. Now, Notebook LM's ability to obviously stay grounded, which is, number one, the most important thing when working with a large language model. But then also now being able to take all of this unstructured data and structure it into tables is extremely huge. And then the last release, which is big, right? And also on the last one, you can export those tables directly to Google Sheets. And then, last but not least, Google promoted some new features that enable you to export and share different creatives from the Studio panel, enabling one click export of study guides, briefing documents and save notes to Google Docs or sheets with full table preservation for handoff or collaboration. All right, yeah, big updates from NotebookLM. Check it out and let me know in the comments, right? Whether you're listening on the podcast, you know, I go back and read the Spotify comments, or if you're listening here on the live stream, which One of those NotebookLM updates are you most excited about? Maybe we'll have to do another, you know, NotebookLM episode, or who knows, maybe we'll do a Notebook LM course on that thing that we're cooking up that you got to pay attention to. We're going to be announcing some details soon. All right, last big news, there is a new AI image model King in town, and it's not named nano banana, so OpenAI has released their new AI image model, GPT5 Image 1.5, their new flagship image generation model that's now available in ChatGPT and the OpenAI API, aimed at generating better images and editing photos more quickly. So according to OpenAI, the model produces images up to four times faster than the previous GPT image model and preserves fine details and user intents more reliably than prior versions. So the update adds a new image tab as well in the ChatGPT app and the browser@chatgpt.com images, which combines an image creation workspace with a discovery and brainstorming feed to inspire users. Which is pretty cool. We see we've seen a similar output in Google Gemini with the My Stuff feature, but a little different, right? It's almost like a little collaborative image workspace in OpenAI, which I thought was pretty cool. But here's the interesting part. So early benchmarks show that GPT5 image 1.5 tops LM Arena's text image leaderboard, where it displaced Google's Nano Banana Pro from first place. So real world edits and reviews so far are mixed, kind of directs comparisons of, you know, different outputs show different things. And I've been doing a lot of outputs myself. So when I saw the LM arena scores at first I was like, I don't think so, right? At least in my limited use up to that point, I really thought GPT Image 1.5 was not quite as good as Google's Nano Banana Pro. But the more that I looked at it and iterated, I was like, okay, I can at least see how they're on the same playing field at least, right? So I don't know if I would necessarily say that GPT image 1.5 is better for all use cases, but on the blind taste test that is LM arena, it is ahead, right? So it also depends on what your taste is, I do think. Right? So even on the same text image prompts, I think a lot of times Nano Banana Pro creates something that looks more real, right? Like it was taken with, you know, a high end smartphone Where GPT Image 1.5 looks a little more polished, right? It looks almost like it's a studio setup, right? It's almost like, you know, skin tones are a little glossy, so it almost looks a little hyper realistic. It still looks really good in leaps and bounds, better than their first GPT image. So I guess it depends what you're going for. One other thing that I will say GPT image 1.5 is probably a little bit better than Nanobanana Pro at is just iterating and understanding the context. I still think Nano Banana is much better at infographic, better at complex images that require a lot of text. I don't think GPT5 image or GPT 1.5 image is there yet, but I think when it comes to iterating and understanding nuanced context of a conversation, I think GPT 5 GPT 1.5 image from ChatGPT. Yeah, it might be a little bit better if you're trying to one shot. If you're trying to get photos that just look right, especially photos of people or groups of people or venues, event spaces. I think Nano Banana Pro still takes it. But hey, it's at least a competition, which means in the end we all win. All right, that is not all of the AI news. That is our big news stories. But let's go ahead and recap the what's new and what's next. So on other weeks, some of these might have been some of our top AI news stories, right? For the most part, each week we give you the top eight to 10 and then we give you a quick bullet point kind of firestorm run through of everything else. So some of these are new product releases, partnerships that have already been confirmed, some are rumors and leaks. But let's get straight into them because this week we actually have a ton. All right, here we go. So OpenAI launched a new real time and text to speech model for devs on its playground. Lovable. The AI vibe coding platform raised $333 million at a $6.6 billion valuation. Chat GPT is rolling out and has rolled out a new writing blocks feature that makes it easier to collab with the system for writing emails and easier to format. Spanish banking giant BBVA rolled out Chat GPT Enterprise to 120,000 staff members, making it one of the largest AI deployments ever. Former presidential candidate Bernie Sanders urged a nationwide halt to new AI data center construction to let democracy, quote unquote catch up. Google updated its Translate with Google or sorry with Gemini AI adding live headphone translation beta for over 70 languages. Really cool. Allows people with using the service to speak in different languages and hear and understand each other. Mozilla hired a new CEO that pledged it would turn Firefox into a modern AI browser. Chat GPT rolled out some new personalization settings which I think so far are amazing, where you can toggle the amount of warmth, enthusiasm, headers, lists and emojis. So there's essentially, you know, default vault less or More for those four different categories. I'm loving it so far. Google has released an open source function Gemma model that's meant to be fine tuned for your specific function calling tasks. Google also unveiled new AI smart glasses with Samsung Gentle Monster and Warby Parker. More Google Google engineers, like I said, hinted that there will be new versions of Gemini 3 that may come out in the next couple of weeks. Cursor acquired Graphite, an AI code review platform. ChatGPT rolled out pin Chats so you can add certain chats as your favorite to the sidebar. Meta added a Sora Cameo esque style feature that lets you add yourself to any AI generated photos or videos. Google announced an upgrade to Deep Research. So now there's visuals in there so you don't have to just stare at, you know, huge blocks of text. It will create visuals in there as well. OpenAI released GPT5 to Codex, its best agentic coding model. Mistral released Mistral OCR3, the best AI OCR model in the world, which is actually a big deal, right? If you're, you know, trying to work with a large amount of PDFs and using AI, you might want to look at Mistral OCR3, a leak anthropic is planning a new agent mode with tasks on the front end, right. So almost having a front end agent mode, that will be pretty cool. Google has rolled out Gems from Labs, which essentially brings in the opal app experience into reusable gems inside Google Gemini. Google also launched cc, an agent that delivers a summary of your inbox. But right now it's super limited and it's a wait list and I believe only personal Gmails can sign up. Google also launched Gen Tabs, a beta feature that turns open tabs into to apps, which is really cool. And then last but not least, Amazon shuffled their AI leadership, appointing Peter Desantis to oversee some new sections including Amazon's AI models, chips, robotics and quantum computing. That was a ton, y'. All. If you think, if you think you can just take the first, you know, the last couple of weeks of December off and, and, and have a slow start in 2026, I'm letting is not the case. Right. Your competitors are still keeping up. The AI companies aren't stopping shipping. Right. If you take a month off and decide to hibernate, you come back, you're going to be behind. I'm letting you know that because yeah, we're taking a, you know, like I said, three days off, you know, this week to recharge. But I'm letting you know, we're going to be launching something very soon. So if one of your big goals personally for your company, for your department is to really invest in understanding large language models and and using front end large language models for your business, trust me, you're going to want to be tuning in. We're going to be releasing some new offerings pretty soon for free, obviously, so don't miss that. It's going to be limited at first as we roll them out. So you got to be reading our newsletter, you got to be opening it and you got to be tuning in. So thank you for listening today if this was helpful. If you're listening on the podcast, please leave us a rating and follow the show. We really appreciate that. Like I said, make sure to check out the 2025 roadmap Rewind episode 674 and 676 from last week. Like I said, we're going to be taking Wednesday through Friday off here to recharge to work on some cool things. That's it. Thank you for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks y'. All.
- [34:22]
A
And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit your everydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.