HUGE Gemini 4 Argon UPDATE, Fable 5.5 LEAKS, GPT-6.1 Ultrafast, Claude Motion, & Step 5! AI NEWS

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: HUGE Gemini 4 Argon UPDATE, Fable 5.5 LEAKS, GPT-6.1 Ultrafast, Claude Motion, & Step 5! AI NEWS · Channel: WorldofAI · 26:46 · Watch on YouTube

Topics: AI agents, Task and workflow automation, Software engineering with AI, AI models, Health and science

Summary

The video covers a range of AI announcements and leaks, including Google’s delayed Gemini 4 Argon, its workplace AI agent, and a possible paid-tier model lineup. It also highlights Step 5 Preview, an open-weight coding model available to try for free, alongside OpenAI’s GPT 6.1 Soul Ultra Fast and interactive ChatGPT UI. Anthropic’s Claude Dashboards and Motion enter beta, while Sonnet 5.5 receives lower prompt-cache pricing. The video ends with an unconfirmed report that a developer used AI coding agents to identify two possible exoplanets, with follow-up observations scheduled for November.

Key points

Gemini 4 Argon did not launch as expected; Google described its ability to process unstructured information and carry out complex enterprise workflows, while broader availability remains uncertain.
Leaked app information suggests Gemini 4 access may vary by subscription tier, with Argon reserved for Ultra users; Google has not confirmed this lineup.
Google introduced Gemini as a workplace agent that can handle tasks across Google and third-party services, create specialized subagents, and use models from different providers.
Step 5 Preview is a 600-billion-parameter mixture-of-experts model with 27 billion active parameters and a one-million-token context window; the presenter says it is temporarily available to try for free on several platforms.
GPT 6.1 Soul Ultra Fast is rolling out with a claimed speed increase of up to eight times over standard Soul, but its API pricing is substantially higher.
ChatGPT’s new intelligent UI can turn responses into interactive visual experiences, including charts, explanations, buttons, and forms.
Claude Dashboards and Claude Motion entered beta; Motion creates editable animations using code rather than an AI video-generation model.
Anthropic cut Sonnet 5.5 prompt-cache pricing in half, to 10 cents per million input tokens, according to the video.
A developer’s analysis using Claude Code and OpenAI Codex identified two possible exoplanets, but the findings are unconfirmed and await follow-up observations in November.

Chapters

Start	Chapter	Summary
00:00	AI news roundup and Gemini 4 Argon delay	The host previews major announcements from Google, OpenAI, Anthropic, and Step, including the expected but delayed Gemini 4 Argon.
04:15	Gemini 4 Argon leaks and availability	The video discusses Argon’s leaked specifications, Google’s comments on its capabilities, and reports that access to Gemini models may depend on subscription tier.
09:07	Google’s universal workplace agent	Google’s Gemini agent is presented as a tool for delegating complex work across apps, with specialized subagents and support for third-party models.
11:17	Fable 5.5 leaks and Step 5 Preview	The host covers reports of a more capable Fable model and compares Step 5 Preview’s coding and 3D-generation results with other models.
16:36	OpenAI’s fast model and interactive ChatGPT UI	GPT 6.1 Soul Ultra Fast arrives with a premium price, while ChatGPT gains real-time steering and automatically generated interactive interfaces.
20:48	Claude Dashboards, Motion, and pricing	Anthropic introduces beta tools for interactive dashboards and code-generated animations, and reduces Sonnet 5.5 prompt-cache pricing.
24:03	AI-assisted search for possible exoplanets	The host describes a developer’s AI-assisted analysis that may have found two exoplanets, while emphasizing that observations in November will test the claims.
26:14	Closing remarks	The host reflects on how coding agents may aid scientific research and closes with channel and community announcements.

Quotes

Google is working to bring the model to more people as safely as possible and as soon as possible — 06:25

Gemini won’t be tied to Google’s own AI models. — 10:46

step five preview got everything right in the first attempt — 15:33

Instead, it writes everything as code and generates that video based off the code that it wrote. — 21:19

these planets that he had discovered haven’t actually been confirmed yet. — 25:40

Concepts and entities

Gemini 4 Argon, Gemini 4 Flash, Gemini, Step 5 Preview, GPT-6.1 Soul Ultra Fast, Claude Motion, Sonnet 5.5, Claude Code, Codex, Pavle

Full transcript (original language)

00:00 Pretty interesting day in the AI space with tons of new models releases as well as major updates from OpenAI, Enthropic, and Google. Starting off with Google as they just held their Gemini at work 2026 event, introducing a bunch of new features, introducing Gemini agent, a universal AI agent that can handle knowledge work, generate images, media, write, and execute code, analyze data, and automate complex workflows. essentially a general AI agent. But what

00:32 we’re really interested in is Gemini 4 Argon, which was supposed to release today, but ended up getting delayed. The reason why we thought it was releasing today is because of the model strings which appeared inside anti-gravity, but it looks like we’ll have to wait a bit longer. Maybe tomorrow, maybe later this week, or who knows. However, we did learn quite a bit about this model. Google CEO Sundar Pichai had highlighted Argon ability to process massive amounts of unstructured information and execute complex enterprise workflows from start

01:04 to finish. More importantly, new information is suggesting that Gemini 4 Argon could be exclusive to Ultra subscribers as well as paid API users. Whereas with Gemini 4 Flash, which is a model that we also saw today, could be restricted to pro subscribers, leaving free users with Flashlight. And honestly, that’s pretty disappointing if true. Google had reportedly moved deep think access back to Ultra in its documentation, and we had also spotted Gemini 4 Argon inside anti-gravity CLI. But the capabilities involved data

01:38 synchronization, building pipelines, as well as production deployments. We’ve also got interesting demos and additional announcements from Google’s event that we’re going to be covering later on. But moving over to step one, which had introduced step five preview, their new flagship model that’s designed for a chance at coding and professional knowledge work. It’s a 600 billion parameter model with 27 billion active parameters and it has a 1 million token context window. But the best part and why I’m covering it is because it is

02:08 something that is free for the week and you can access it across multiple platforms like open code and open router. Over at OpenAI, GPT 6.1 soul ultra fast is rolling out today cross API codecs chat GBT work and promising near astro level intelligence at speeds up to eight times faster than standard soul. OpenAI also introduced a pretty incredible chat GBT update called intelligent UI, which is transforming ordinary text responses into visual

02:39 interactive experiences, making complex topics easier to understand and allowing you to interact with tools directly inside your conversations. Personally, I think this is one of the coolest CHAGT updates we have seen in a while. Meanwhile, Enthropic introduced Claude dashboard and Claude Motion, both entering beta. You can now ask Claude to turn your data into interactive dashboards or transform reports, charts, and product walkthroughs into animated explainers. What’s especially interesting about Claude Motion is that

03:11 it can generate animations entirely through code, not just AI video models. Meaning, you can edit individual elements, adjust timings, and export everything as a MP4, which is actually pretty cool. And that’s not even everything, as we’ve got also a huge Sonic 5.5 pricing update that makes the model an absolute steal, some interesting new Fable 5.5 leaks, and a pretty wild story making headlines about someone allegedly using Claude to discover a new planet that wasn’t even known by NASA. We’re going to be taking

03:44 a closer look at that story along with some impressive Gemini 4 demos and all of the other major announcements. And I’m really sorry about my voice right now. You can tell that it’s a little grimy, but that is because I am a bit sick. But I wanted to get you guys up to date on whatever is happening in the world of AI. So, it would really mean a lot to me if you guys can just please like this video and just subscribe as majority of you guys aren’t actually subscribed. And it really helps me make videos for you guys on a day-to-day basis. But with that thought, let’s just simply dive into today’s upload. This

04:15 villa, this game, both built with AI. This is only the beginning. Meet Avacus AI supercomput build with GBT6 Astra and Opus 5.5 right from a prompt. Design a 3D space you can explore. Create a game you can actually play. Then keep refining it through chat. Astro mode brings your game ideas to life from dark fantasy worlds to racing circuits with tools for modeling physics and testing.

04:45 And when you’re building apps, go beyond the visuals. Add a database. User signin and Stripe payments with free tiers for app hosting and databases. Bring the idea, shape the experience. Build something people can explore, play, and use. All in Abacus AI supercomputer. Thanks to Abacus for sponsoring today’s video. Your next idea deserves more than a prompt. So start building with links in the description. Let’s start things off with Google and what was revealed about Gemini 4 Argon during today’s Gemini at Work event. Now, we genuinely

05:19 thought that this puzzle was actually going to launch today, especially after Gemini 4 Argon was spotted inside Anti-gravity’s internal client code just hours before the event. And the leak details were pretty interesting. We’re talking about a dedicated reasoning effort, a context window control system with 512K and 900K context options, as well as quota multipliers of 1.3 times as well as 1.8 eight times and even closed pricing listed at $4 per 1 million input tokens and $20 per 1 million output tokens. So naturally,

05:52 everyone thought that the release was imminent. But Google CEO at the event gave everyone an update. And according to Pitch Eye, early testers have been giving great feedback about Argon’s ability to process massive amounts of unstructured information into cohesive outputs and execute complex enterprise workflows from start to finish. But here’s where things get interesting. Vai also described Gemini 4 Argon as an extremely capable frontier model with no safety safeguards, revealing that Google has been sharing it with a select group

06:25 of cyber security defenders through its Fairon program. He went on to say that Google is working to bring the model to more people as safely as possible and as soon as possible, which essentially means we’re still waiting for a broader release despite all the signs pointing towards a launch today. And honestly, I’m starting to wonder what’s going on behind the scenes. Is Gemini 4 Argon potentially operating at the same level as Enthropics Mythos with capabilities powerful enough to warrant a more cautious rollup? Well, obviously that’s

06:56 speculation and we don’t know whether safety concerns are actually going to be responsible for the delay. But between the leaked specifications, Google’s comments about cyber security safeguards, and the continued uncertainty are surrounding this release, it’s getting increasingly interesting. And we did get a couple of glimpse of Gemini 4 Argon actually functioning at the event where it was actually seen within anti-gravity CLI handling data syncs building out pipelines and it is something that did quite well with production deployments on command and it did it quite

07:30 thoroughly. Now take this with a grain of salt but unfortunately we’ve got some pretty disappointing news about Google’s upcoming model lineup. But we do have one good news. We have a Gemini 4 flash model that was revealed within the backend logs that we’re going to be talking about in a bit. But there was a notification that was spotted inside Gemini’s app that revealed that Flash and Pro models will soon be exclusive to paid plants. This is going to leave free users with unlimited access to only the flashlight model. And based on the

08:00 information that’s circulating, here is how the Gemini 4 lineup will be structured. Free users will only get access to flashlight. Only AI plus users will get flashlight and flash. AI Pro will get flashlight, flash, and pro. And then AI Ultra will get all of that, including the new Argon model. Essentially, this could mean that Gemini 4 Flash requires a paid subscription, while Gemini 4 Argon is locked behind Google’s most expensive Ultra plan for the time being and soon will be deployed to every prouser. Now, I’m not too sure

08:35 if this is going to be what they end up rolling out with, but that is what was seen within the Gemini app, and that if it is true, it is going to be really bad. If you want the best AI tools, workflows, and drops before everyone else, join my free newsletter with the link in the description below, which is completely free. But heading over to the good news, Gemini 4 flash preview is a model string that was found inside Google’s Geni codebase, which is another promising sign that this model could be also coming alongside Argon. This would

09:07 be the flash tier, a part of that model series, which means that it is going to be more costefficient as well as something that is very focused on speed. And honestly, it is crazy that we’re seeing roughly three to four flash model releases or iterations in just 3 months, which is just an insane pace of development, which also adds the fuel in the whole RSI theory surrounding Google. But yeah, let me know what you guys think about all of that. But there was a lot more updates coming from the Gemini at work 2026 event with the biggest one

09:41 being Gemini is evolving into a universal AI agent for work. Essentially, Gemini can now handle everything from research and data analysis to generating images, creating presentations, writing code, and executing complex workflows. So, instead of giving it step-by-step instructions, you can simply delegate a task and let it handle the work. And it’s kind of crazy. We’re starting to see every AI lab develop their own generalist AI agent. We saw it first with all of these smaller companies like Muse, which was

10:13 originally built on Manis. Then we saw Grockbot and now we have OpenAI’s thoughts. But this is where Google is now creating their own universal general AI agent. And what’s really interesting is that this same agent works across Gmail, Docs, Sheets, Drive, Slack, as well as Microsoft 365 and even your command line, all while sharing the same memory, context, and personalization. Google also has introduced the ability for Gemini to dynamically create specialized sub agents. This is to handle different parts of complex tasks

10:46 alongside persistent co-worker agents that function more like actual team members. And on top of that, we’re getting new natural language data analytic tools, which is going to be industry specific agents for finance and legal work and centralized security controls for businesses. But here’s arguably the most interesting part. Gemini won’t be tied to Google’s own AI models. This is where businesses can choose between Google models, thirdparty models like Opus as well as Sonnet, which is kind of mind-boggling, and even

11:17 open- source alternatives depending on the task. So, Google is essentially trying to turn Gemini into an entire AI workforce rather than just another chatbot. Next up, we have a pretty interesting development surrounding the Cloud Fable 5.5, which could be right around the corner. Now, people have reportedly noticed that some Fable 5.1 requests are appearing to be rerouted to a more capable model with noticeable improvements in performance, which is likely the new Fable 5.5. And

11:49 considering Enthropic has already released other models in this 5.5 lineup, we just saw the 5.5 Hiku, it wouldn’t be surprising if Fable 5.5 is next. Nothing officially confirmed yet, so just give that with a grain of salt. But this is insane, guys. This is Fable 5.5 fully creating a fullon Minecraft clone, which is better than everything that we have seen. Exactly to the tea. It was able to replicate all the textures, all the functions, and this is

12:20 just insane to me. I’ll leave a link to this, but we’re talking about a fully generated UI of the original Minecraft game. the graphics, the mobs, you have different biomes, caves, items, and gameplay mechanics, all from a single prompt coded up by this new Fable 5.5 model. And honestly, the quality is so much better than what we have seen from Astra as well as Opus. So, it’s insane that Fable 5.5 is going to be at this level and it’s going to be able to

12:52 output this sort of generation for anything that you do across multiple domains. If you like this video and would love to support the channel, you can consider donating to my channel through the super thanks option below. Or you can consider joining our private Discord where you can access multiple subscriptions to different AI tools for free on a monthly basis, plus daily AI news and exclusive content, plus a lot more. Moving forward, we have another pretty impressive new AI model from Step One called Step Five Preview. their

13:24 latest flagship model designed for genensic coding and complex professional workloads. And the specifications are pretty crazy because we’re looking at a 600 billion total parameter mixture of experts model with just 27 billion active parameters. It has the 1 million token context window with vision capabilities. And Stefun claims that it delivers frontier level performance at significantly lower costs with particular strengths in software engineering, finance, and longunning autonomous workflows. Now, in fact, they’ve had tested this model on tasks running for up to 24 hours, including

13:59 GPU kernel and optimization and automated model post training where it can continuously evaluate results, learn from the execution feedback, and improve its approach. And the model weights are scheduled to be released in a couple of days on October 15th. But the main reason why I’m covering this model is because you can actually try it completely for free right now for a full week across different platforms like Open Router, Hilo Code, Klein, Open Code, and a couple of others like news research. So I would highly recommend

14:31 making the most out of it cuz this is a pretty solid model that you can use completely for free right now. And to be fair, when you are to compare this model with similar model sizes as well as other openweight models like Deep Seek version 4, you have GLM 5.3 and Quen 3.8 on the same identical 3GS coding task. This is where it was requested to build the Eiffel Tower simulating a rocket launch and creating a realistic building collapse. This is something where you would assume that step five doesn’t actually keep up with these other open

15:02 models, but surprisingly step five preview was both the fastest and cheapest across all three tasks while still delivering some pretty solid visual results. You can argue that it did a better job than all three of the other models. But this is just insane that we’ve got a new open model that is capable of outputting these sorts of generations, especially in the domain of coding. I may be doing a full dedicated video on this later on, but this is another interesting comparison between Step Five Preview and Deepseek version 4

15:33 Pro where both models were given the exact same task to build out a 3D Voxil style Japanese garden and 3JS complete with a pagota as well as with cherry blossoms and a koi pond. Both models use the same prompt, fresh sessions, and medium reasoning settings. And surprisingly, step five preview got everything right in the first attempt, completing the task in just 35 minutes. Now, usually these are prompts or requests that take a longer time, but this is something that got it done relatively cheap. Now, you will notice

16:05 that there are a couple of errors, but that is obviously adding a couple of things that you can refine afterwards, but Deep Seek version 4 Pro also had those errors as well and needed three rounds of fixes and took 43 minutes to finish. That is a pretty impressive result for a step five, especially considering it’s an openweight model. And again, you can try it yourself through Open Router, Hilo Code, and all those other apps that I had mentioned completely for free right now. So, I’m not sponsored by these guys, but I would

16:36 highly make the most out of it for this small period. Next up, a part of the 28day pledge to ship new features at OpenAI. Chad GBT has gotten an upgrade. This is where OpenAI is going to be now releasing the GBT 6.1 soul ultra fast which is rolling out obviously within chat GBT but through the API codeex as well as chat GBT work and OpenAI is making some pretty serious claims here promising near Astro level intelligence at speeds up to eight times faster than

17:07 standard GBT 6.1 soul that’s genuinely impressive especially for coding agents debugging production issues as well as other workflows where you need every second. But here’s the catch. The pricing is really ridiculous. We’re looking at $12 for 1 million input tokens and $60 for 1 million output tokens through the API. And honestly, that’s expensive and it’s basically the same pricing as Astra if you think about it. Especially considering that the standard GBT 6.1 Soul is already significantly cheaper. You’re

17:40 essentially paying a massive premium for faster inference. And I’m not convinced that makes sense for most developers. So yeah, you could be running time-sensitive production workloads where speed is directly translating into money, then I can see that this is something that would kind of resonate with me. But for everyday coding tasks, I think standard soul offers much better value. And obviously at the current moment, I personally believe that OpenAI is not competing as well with the rate limits as what Enthropic is providing

18:11 throughout cloud code as well as their API. Now, as per availability, Ultrafast is currently accessible through the Pro $500 plan alongside eligible usagebased enterprise and credit based EDU plans with administrators needing to enable access for enterprise users. OpenAI is also expanding data residency support including EU residency for GBT 6.1 Soulfast and GBT6 Luna Fest. So, overall, this is a pretty decent speed improvement, but I don’t know if the

18:42 pricing makes sense. And alongside ultra fast, OpenAI also has made a pretty useful improvement to GPT 6.1 Soul’s steering capabilities. According to OpenAI’s Tibbo, steering is now essentially instant, meaning you can redirect the model in real time while it is working without waiting for it to finish. And it is something that does kind of good with its approach. So if your coding agent is heading in the wrong direction, you can immediately course correct it instead of wasting time and tokens on something you didn’t

19:12 want. Now, this is a part of yesterday’s release, but I didn’t actually make a news video yesterday, so I wanted to cover it today. And essentially, this is a pretty impressive new feature added to Chat GBT, and that is the intelligent UI that is powered by GBT6, and this is something that is rolling out to everyone. Essentially, Chat GBT is moving beyond traditional textbased responses within ChatGpt, and it can now generate fully interactive experiences directly inside your conversation. and I love it. This is where we’re talking

19:44 about dynamic charts, visual explanations. You have interactive buttons, forms, and even custom interfaces that adapt to whatever you’re asking. And instead of just reading through a wall of text previously, this is where you can actually interact with the information and you can even explore different options and understand complex topics visually. So this is something that they have introduced which is really cool because GBT6 can automatically decide when to use these interactive elements based on your

20:15 question without you having to specifically request them. Personally I think this is one of the best updates OpenAI has made in a long time to chat GBT and it is something that is essentially going to move forward a future where AI doesn’t just need to answer your questions and just output a wall of text. This is where it can help you, work with you, and it’s going to be able to even use its capabilities to help you with these visual elements. Moving back to Enthropic, they’ve also announced some pretty exciting new features for Claude. And it kind of makes sense that they released these

20:48 features right after what Openey did with their chatbot updates with the element UIs because this is where Enthropic is releasing Cloud Dashboards and Cloud Motion both entering beta. Starting off with claude dashboards, you can now connect their data platforms or CRM tools to simply ask claude questions in natural language. This is essentially where Cloud can automatically write the necessary queries and then it can generate interactive dashboards that update as your data changes. You can even inspect the underlying queries

21:19 behind each chat, which is pretty useful. And this is currently rolling out in beta on all paid plans. But what’s arguably more interesting is Claude Motion. This is something I personally like because it lets you turn reports, charts, and product walkthroughs into polished animated videos. And here’s the cool part. Pot doesn’t use the AI video generation model to create these animations. Instead, it writes everything as code and generates that video based off the code that it wrote. That means you can easily change the text, numbers, animations, and timings without

21:53 regenerating the entire video and export the final result as an MP4. This is also something that is currently in beta both for team and enterprise subscribers and I can definitely see this being a useful feature that is going to be used for presentations, educational content, and even product demos. Lastly, cloud docs, slides, and design are officially out of beta with all three now available across every Cloud plan, including the free one. You can even collaborate with your team and claude on the same document, presentation, and design. So, this is

22:27 honestly a great list of new updates that the team has introduced and it’s basically making Claude as interactive as what OpenAI did. Obviously, not at the same extent, but it’s a way for them to rebalance the shift. Now, I’m just reiterating on this just in case if people haven’t actually seen this, but essentially when Haiku 5.5 dropped, there was also an update for Cloud Sonic 5.5 in regards to a significant pricing update, which honestly makes this model

22:58 even a better deal and a better model to use. Enthropic has officially cut the price of Sonnet 5.5’s prompt cash needs in half. This is bringing it down to just 10 cents per million input tokens. For context, regular input tokens will cost about $2 per million input tokens, while output costs about $10 uh dollars per million. And according to Enthropic, this changes that and makes Sonet 5.5 roughly 20% cheaper for most Agentic workflows. It’s especially a big deal if

23:29 you’re going to be coding for long sessions, if you’re repeatedly reusing large amounts of context, where caching can save you a ton of money and even tokens. So, this is another huge update from Enthropic, and I don’t know how they just keep doing it. I believe they’re releasing all of these features, all of these models in preparation for their IPO, which is going to be coming maybe in November. Sorry about this, but back to Google. There’s another quick reminder I should have mentioned before, but this is only for anyone who subscribed to Google AI Pro and Ultra. You can actually claim your monthly

24:03 Gemini API credits which you can then use across Google supported AI models. And the best part is is that these credits aren’t limited to the Gemini app. You can use them through your own API key in your own projects, anti-gravity CLI, or compatible thirdparty coding agents and harnesses. So, if you do have these plans, I would highly recommend making the most out of it cuz you’re essentially getting these free credits. And lastly, something that blew me away is a pretty wild story about a developer who may have

24:34 discovered two previously unknown exoplanets using Cloud Code and OpenAI’s codeex. This is where a developer named Pavle had spent two weeks analyzing astronomical dates with AI coding agents, running 74 separate analysis and over 100 scripts to investigate unusual dimming patterns. And this is from a star that is called TIC 4206066. And the results were pretty fascinating. He even tested its findings by hiding individual years of observations and

25:07 seeing whether the remaining data could predict when that star would be dim. The predictions were correct three out of three times. He then rebuilt his entire analysis from scratch, ruled out unknown nearby stars that could explain the signal, and search through over 340K automated alerts from NASA’s planet hunting software. And apparently, neither potential planet had been previously identified. But here’s where things get even interesting. NASA’s test mission approved his proposal for follow-up observations, meaning the

25:40 telescope is scheduled to observe the star again in November, collecting measurements every 2 minutes. Pabell has even published his predicted dimming times in advance, so when the new observations arrive, anyone can independently check whether his predictions were accurate. And I’m definitely going to keep you guys posted on it cuz this is honestly an insane story. Now, to be clear, these planets that he had discovered haven’t actually been confirmed yet. The November observations will be an important test for this finding. And honestly, this is incredible cuz it just showcases of how

26:14 AI coding agents could help accelerate real scientific research, even for cases like this. We’re talking about one developer using Claude Code and Codeex to conduct an investigation that could potentially lead to a discovery of two new planets. But that’s about it guys for today’s video. Sorry for the weird voice or yeah, you can say weird voice. I have been sick. So I tried my best to make this video for you guys so that you guys can stay up to date with the latest AI news. But with that thought guys, I hope you enjoyed today’s video. Make sure you go ahead and take a look at the world of

26:46 AI school community. Take a look at the universe of AI which is our second channel. Take a look at our vibe coding platform and benchmark as well as our newsletter, Discord, follow me on Twitter. And lastly, make sure you guys subscribe, turn on notification bell, like this video, and please take a look at our previous videos so that you can stay up to date with the latest AI news. But with that thought, guys, have an amazing day, spread positivity, and I’ll see you guys fairly shortly. He suffers.