The rumors have hit a high pitch since Altman teased the model in a podcast two weeks ago, and various aliases for advanced models started showing up at some benchmarking sites.
It will also run longer than usual, as there is a lot to cover, says @sama:
Potentially all of government civilian employees are covered by the deal, reaching some 2—3 million in the current federal workforce. That would be a boon for OpenAI, but more in the sense of prestige than in usage, as ChatGPT currently has around 700 million weekly users.
The agreement could likely cement a full year of entrenched habits, history and preferred use if successful — basically locking in OpenAI as the preferred AI provider across federal agencies.
Continue reading “OpenAI giveaway: All of government to get ChatGPT for a year for $1 per agency”
Sam Altman says «We believe far more good than bad will come from it,» choosing to release the models after a series of delays and worry about the weights.
After «billions of dollars of research» and extensive red-team testing, they were found no more dangerous than the o3-model and won’t move the needle on chemistry or biology.
Continue reading “Gpt-oss: OpenAI releases open weights model after extensive testing”
That number spans all of ChatGPT’s accounts, ranging from free to Pro and everything in between, and is rapidly accelerating, writes CNBC.
Continue reading “OpenAI says 700 million users this week, announces wellness updates”
ChatGPT was just two clicks away from spilling your secrets to Google, an investigation found yesterday. The «Make this chat discoverable» button on the share feature would register the whole chat on search engines. Plenty of people made that mistake, sharing «deeply personal details, including struggles with addiction, experiences of physical abuse, or serious mental health issues.» OpenAI removed the feature shortly after, saying it «introduced too many opportunities for folks to accidentally share things they didn’t intend to.» They are now scrambling to remove «indexed content from the relevant search engines.»
Apple open to mergers and acquisitions in AI space
— We’re very open to M&A that accelerates our roadmap, Apple’s Tim Cook said on yesterday’s earnings call. — We are not stuck on a certain size company, although the ones that we have acquired thus far this year are small in nature. He also said Apple was going to «significantly» grow it’s AI investments, after the company reported a 10% increase in revenue — the largest jump since 2021. They are also «making good progress» on personalizing Siri, he said.
Zuckerberg throws shade on open source projects
Mark Zuckerberg of Meta recently posted a manifesto of sorts, mapping out what he sees as a benevolent «personal superintelligence» «for everyone.» In it, he quietly states that superintelligence will pose new safety concerns, and «We’ll need to be rigorous about mitigating these risks and careful about what we choose to open source.» He repeated this in his later earnings call, saying «we kind of wrestle with whether it’s productive or helpful to share that.» See also his original post.
Developers are slightly souring on AI coding
A new survey from Stack Overflow shows a significant drop in developer trust in AI coding, with trust in its accuracy dropping from 40% in previous years to 29% in 2025. «Favorability» has also dropped from 72% to 60% on a yearly basis. 52% of developers say they use AI agents in their work, while 72% reject «vibe coding.» The survey was taken with 49,000 worldwide developers. Stack Overflow is no stranger to AI effects, having dropped sharply after the AI coding boom.
It will run mostly ChatGPT loads, but «surplus capacity» will be made available to local and North European developers, and government programs, which is in high demand.
ChatGPT use in Norway has quadrupled in the past year, OpenAI says in its statement on the project, and there are thousands of developers on the platform.
Continue reading “OpenAI unveils Stargate Norway — its first data center in Europe”
The general idea behind the mode is that it can ask Socratic questions to assess your knowledge levels and guide you step by step through problems — to help you learn instead of just giving you a quick fix.
Continue reading “OpenAI debuts «Study Mode» to help students learn, not just copy”
After saying at the model launch that they were still working on European access, less than a week later, OpenAI sends this tweet:
Pro users in the European Economic Area and Switzerland, thanks for your patience — ChatGPT agent is now fully rolled out to you.
And the rollout to Plus users globally has started and will continue over the next few days.
— OpenAI (@OpenAI) July 23, 2025
That clearly says the agent is coming to Pro Users in the «European Economic Area and Switzerland.»
It also says Plus users — globally — should have access over the next few days.
So if you happen to be in Europe, are a Pro or Plus subscriber, you should get access to the model shortly.
UPDATE 25/7/25: OpenAI reports that the rollout is complete. You can find it in the feature bar at the bottom of the app window by clicking on the “Agent” icon.
Their new 4.5 GW data center in Abilene, Texas, built in cooperation with Oracle, is indeed part of the Stargate project, they say. As a reply to the report, they are showing operations are already starting.
Continue reading “Stargate was not dead: OpenAI announces 4.5 GW data center in Texas”
But here’s the truth: None of these numbers are easy to verify — and each company counts differently.
Continue reading “So, who really has the most GPUs in the AI race?”
OpenAI says it was achieved with «our latest experimental reasoning LLM:»
Continue reading “OpenAI wins Math Olympiad gold, a major milestone for AI development”
The ChatGPT Agent should be available in the model selector at the bottom of the ChatGPT window in the app for Pro, Plus and Teams users as of Thursday, but rollout will be «over the next few days,» or over «the coming weeks» for Enterprise and Education users.
Continue reading “OpenAI launches ChatGPT Agent mode, for tasks both easy and tough”
OpenAI is pushing a livestream tonight/early morning at 10am PST/1900 CET, with a tweet saying:
ChatGPT 🤝 Deep research 🤝 Operator
Livestream in 3 hours.
— OpenAI (@OpenAI) July 17, 2025
Some expect the company to launch Odyssey — a rumored blend of deep research, agentic capabilities, and access to your Drive files, and they seem on the money. If true, it could revolutionize coding and research workflows.
Others are fixated on the cursor in the teaser, which circles through five points — surely that must mean ChatGPT-5? Meanwhile, another camp is betting on the long-awaited OpenAI browser, complete with built-in agents.
Speculation is running hot. The only way to know for sure? Tune in later — or tune out and catch the fallout in the news.
Here’s the teaser posted last night, you can likely catch it on Youtube, and keep an eye on OpenAIs x.com account.
UPDATE: it was ChatGPT Agent. The news is here.
OpenAI had previously planned for June release, and then postponed it to «later this summer,» claiming great progress and that the team was «excited» about the new model.
Very good, or just recycled o-series?
It was supposed to be a «very good model,» but «our research team did something unexpected and quite amazing,» @sama said on June 11.
Continue reading “OpenAI gets hit with open-weights reality, delays model indefinitely”
The deal will see Windsurf CEO Varun Mohan, co-founder Douglas Chen and a small team join Google’s DeepMind division.
Also licensing key tech
Further, Google will invest $2.4 billion in a non-exclusive deal to license Windsurf’s technology, reports Reuters, among others.
OpenAI had been in long-winding talks to buy the company, in its biggest deal yet, and many said it was just around the corner, as late as in May, 2025.
Continue reading “Google hires top execs, team from Windsurf — upending OpenAIs deal”