I just added real-time AI translation into Chormex using GPT-Realtime-2… and this feels absolutely surreal. It works across YouTube videos, live streams, meetings, presentations, basically anywhere audio is playing inside Chrome. You can watch translated speech in real time while simultaneously using Codex on top of the live context. “Summarize this.” “What are the key points?” “Turn this into notes.” “Explain what they mean.” “Organize the discussion.” …all while the video or meeting is still happening. It genuinely feels like browsers are evolving into real-time AI operating systems. We are getting dangerously close to a world where language barriers on the internet completely disappear.

When you’re using Codex’s /goal feature, try setting the stopping condition based on time. Every time you finish implementing a unit of functionality, check the current time and just keep going… all the way until 9 AM the next day. And if the clock is ticking down and you haven’t nailed the detailed implementation yet? No panic, just prioritize the most important pieces and push those through first. Basically… turn your AI into an all-night coding gremlin with a deadline

This GPT Image 2 prompt is going insanely viral right now. “Redraw the attached image in the most clumsy, scribbly, and utterly pathetic way possible. Use a white background, and make it look like it was drawn in MS Paint with a mouse. It should be vaguely similar but also not really, kind of matching but also off in a confusing, awkward way, with that low-quality pixel-by-pixel feel that really emphasizes how ridiculously bad it is. Actually, you know what, whatever, just draw it however you want.”

OpenAI will face growing pressure to go public. The company has already made promises to the investment market that are too large to remain private forever. Future value, AGI expectations, data centers, and dedicated compute infrastructure have all become the logic behind the capital it has raised from investors and corporate partners. At the end of March 2026, OpenAI closed a $122 billion funding round at an $852 billion post-money valuation. At that scale, saying "the potential is huge" is no longer enough. The company now has to prove itself in numbers. It has to show how fast users are growing, how many of them can become paying customers, how deeply enterprise customers are adopting the product in real work, and whether revenue can grow fast enough to carry the compute cost behind all of it. That is why I think IPO pressure is becoming an important issue for OpenAI. There will probably still be many investors who want to put money into OpenAI. The question is not whether there is interest. The question is how long the private market can absorb this scale. OpenAI's valuation now looks less like a normal startup valuation and more like the valuation of a national-scale infrastructure company. Data center commitments and GPU contracts have to be decided years in advance. From the investor side, the next demand may no longer be only a higher private valuation. It may be liquidity, public verification, and a clearer view of the economics. In other words, OpenAI is no longer becoming a company that can raise money like a startup. It is becoming a company that has to prove its economics to the public market. Recent reports make this pressure more visible. OpenAI reportedly missed some internal user-growth and revenue targets, and ChatGPT reportedly did not reach the company's goal of one billion weekly users by the end of 2025. On the surface, more than 900 million weekly users is already an enormous number. But for a company approaching the public market, the absolute number is not the only thing that matters. What matters is whether the company can keep the growth pace it has already implied to the market. Reports that CFO Sarah Friar has raised concerns about whether OpenAI can afford large future compute contracts if sales do not grow fast enough point in the same direction. I do not see this only as an internal conflict. It looks more like a signal of where OpenAI's business model has arrived. Revenue is growing, but compute commitments are growing even faster. Users have exploded, but free users still dominate the base. The question eventually becomes simple: can OpenAI turn massive usage into a real revenue structure? The problem is that someone has to fund the time between now and that answer. The models still have to run. Users are still increasing. Competitors are still moving. Until Stargate and other long-term infrastructure bets are fully built out, OpenAI has to buy time with cloud partnerships and private capital. This is why I think an IPO could become a massive financing event that helps OpenAI survive the transition from model company to infrastructure company. The recent restructuring of the Microsoft partnership also makes more sense in this context. Microsoft remains OpenAI's primary cloud partner, but the new structure gives OpenAI more room to make its products and services available across other clouds. Microsoft will continue to have a license to OpenAI IP for models and products through 2032, but that license is now non-exclusive. Microsoft will no longer pay revenue share to OpenAI, while OpenAI will continue paying Microsoft a share of revenue through 2030, subject to a cap, according to AP's report. This does not look like a breakup. It looks more like a relationship being reorganized for an IPO. If OpenAI is too tightly tied to Microsoft, it becomes harder to describe itself to public investors as an independent platform company. For Microsoft, it also makes sense to keep Copilot and Azure's core partnership with OpenAI while reducing the risk of having its entire AI strategy depend on one company. Both sides need less exclusivity and more optionality, not a clean break. If OpenAI wants to enter the public market, it has to be able to say something very specific: we are not a Microsoft subsidiary; we are an AI platform company that can work directly with multiple clouds and enterprise customers. Reports about tension between Sam Altman and CFO Sarah Friar over IPO timing fit into the same frame. Altman reportedly wants an IPO as early as the fourth quarter of 2026, while Friar has reportedly urged more caution because of organizational readiness and the risk of massive spending commitments. This looks like a collision between the speed of vision and the reality of finance. The CEO wants to convert market heat and a huge narrative into a public listing while the window is open. The CFO knows that once the company enters the public market, it will be judged every quarter by numbers. For a company like OpenAI, with enormous server spending and losses that have to be justified as infrastructure investment, it becomes difficult after an IPO to explain everything with the phrase "investment toward AGI." The lawsuit with Elon Musk is also more than legal noise. On April 27, 2026, the trial between Musk and OpenAI began, with Musk arguing that OpenAI had abandoned its original nonprofit mission and moved toward a for-profit structure. AP described the trial as a high-stakes fight between Musk and Altman that could reshape the future of AI development. Recent filings and reporting also suggest that Musk has tried to strengthen the message of the case by saying any damages from a victory should go to OpenAI's nonprofit arm and by seeking changes around Altman's role on the board. I do not think this lawsuit is likely to stop OpenAI's business entirely. But for a company trying to go public, it creates a different kind of problem. IPOs are not only about revenue and growth. They are also about governance and risk. OpenAI's transition from a nonprofit research lab for humanity into a hundreds-of-billions-of-dollars AI platform company is already a difficult story to explain to investors. Musk's lawsuit keeps pressing exactly on that weak point. So OpenAI's current strategy looks fairly clear. It has to distribute tokens aggressively. More precisely, it has to give users generous access and low friction so that usage becomes habit. Resetting or expanding Codex limits, making the product easier to try on free and lower-cost plans, and letting more users feel the product directly can all be read as strategic choices. This is customer acquisition cost paid in compute. At the same time, OpenAI has to turn Codex into something much larger than a coding tool. In April 2026, OpenAI said Codex had passed three million weekly developer users, and two weeks later it said the number had crossed four million, according to reports such as The Economic Times. Codex is also adding computer use, image generation, browser workflows, automations, and previews for documents, spreadsheets, and PDFs. It is moving closer to a work operating system. That matters. If ChatGPT created the mass consumer habit of using AI, Codex can become the path that turns that habit into actual productivity. And that is exactly the transition the IPO market will want to see. The story cannot only be "people use a chatbot for fun." The stronger story is "people are beginning to work through this system, and it is becoming difficult to work without it." OpenAI already has more than 900 million weekly ChatGPT users, but AP reported that Friar said about 95% of them do not pay anything. That number cuts both ways. It proves overwhelming consumer reach, but it also means enormous compute cost. OpenAI has to keep the consumer distribution channel while creating stronger monetization in enterprise, developer, and professional markets. Codex is the bridge between those two worlds. Individual developers can use Codex and feel that the product actually works. Enterprises can use Codex for code review, testing, pull requests, legacy analysis, internal tools, automation, and measurable productivity gains. OpenAI can then turn that usage into an IPO story. The story would sound something like this: we are not just a chatbot company. We are an AI operating company with consumer distribution, enterprise workflows, a developer platform, and our own compute infrastructure. That is probably close to the picture OpenAI wants the market to see. No matter how good the GPT models become, the public market will still ask the same question: how much money can this make? To answer that, OpenAI cannot behave like a company trying to conserve every token in the short term. It may have to behave like a company deliberately spending tokens to win user habit. The important metric right now is not only short-term margin. It is the share of daily behavior. In the platform wars of the AI era, the company that becomes the default interface of everyday life will win. Search did that. The smartphone home screen did that. Messaging apps and browsers did that. I think this is one reason OpenAI is pushing Codex toward a super app that can handle more and more of the computer. Before an IPO, OpenAI needs proof of repeated use. Before an IPO, it does not need only the sentence "AGI is coming." It needs the number that says millions of people are already doing real work with this. Before an IPO, it does not need only a larger dream. It needs a revenue structure that can carry the dream. That is why I expect OpenAI to move more aggressively. It will likely open up Codex usage further, bring Codex deeper into ChatGPT, expand enterprise Codex Labs and system integration partnerships, and teach consumers to see it not only as a chatbot, but as an app that can make things. Internally, OpenAI has to use Stargate and cloud partnerships to deal with compute shortages. Externally, it has to build trust that this is not only a company burning money, but a company that can turn infrastructure into economics. If that works, OpenAI could raise an extraordinary amount of money in an IPO. If it fails, the AI bubble narrative could return with force. An $852 billion valuation, hundreds of billions of dollars in compute commitments, more than 900 million weekly users who mostly do not pay, data centers that inevitably take time to build, the restructuring of Microsoft ties, and governance questions sharpened by the Musk lawsuit can all look like a powerful moat. They can also look like a burden too large to carry. OpenAI's core task is now simple. It has to move from being a company that promises the future to being a company that proves the future in numbers. And I think Codex sits at the center of that transition. Codex is one of the clearest pieces of real-usage evidence OpenAI can show the IPO market. If users grow, work time decreases, enterprise workflows absorb the product, and compute turns into revenue, OpenAI's massive infrastructure bet can be read as an advance payment for future monopoly power. Before OpenAI goes public, it has to prove one thing: we are both a consumer platform and a work platform for the AI era, and this enormous compute investment will eventually become money. Codex is the most realistic front line for proving that.

🚨 Holy shxt… The rules of AI image generation just completely changed. A GPT-based model currently testing in the arena under the bizarre name 'duct-tape' is turning the global AI community upside down. What exactly did they feed this thing? → Native-level text generation with zero awkwardness. → Chilling consistency maintained down to the pixel. → Overwhelming illustration quality ready for immediate commercial use. "Is it just downloading photos from the internet?" That was every tester's first reaction. It's that unbelievable. A game-changer that instantly crumbles Nano Banana Pro's dominance has arrived. A lot of people might need to start packing their desks again. The 10 images below were 100% drawn by AI. See for yourself 🧵