Tag: openai

Peter Steinberger of OpenClaw joins OpenAI, agents to become «core»

OpenClaw, the open agentic platform you can steer from your favorite app to do lots of useful things, went viral over the last the couple of weeks — with millions of views, downloads, and 100k stars on GitHub.

It also spawned Moltbook, and showed the future of intelligent AI agents for all to see — as the founder, Steinberger, was inundated with investment and job offers.

Continue reading “Peter Steinberger of OpenClaw joins OpenAI, agents to become «core»”

Internal OpenAI model aids in physics discovery, answers expert maths quiz

OpenAI contributes to a solution for a physics problem that had long vexed scientists. (Picture: generated)
Having long said that discovery is the next benchmark, the model by OpenAI assisted researchers from the Institute for Advanced Study, Vanderbilt University, Cambridge University and Harvard in determining that gluons in some cases have amplitude, a key finding in quantum research.

The model helped scientists by simplifying expressions and made a simple formula for this general case.

Then it spent nearly 12 hours reasoning the new problem, verifying the formula and «producing a formal proof.»

Continue reading “Internal OpenAI model aids in physics discovery, answers expert maths quiz”

OpenAI introduces Codex-Spark, greatly improving coding speed

Codex-Spark is small, fast, and almost as good as the real thing. (Pictures: OpenAI)
Thanks to their recent collaboration with Cerebras, the new model delivers «more than 1000 tokens per second while remaining highly capable for real-world coding tasks.»

The drawback is that it’s text-only and only has a 128K context window, and it’s supposed to be used «where latency matters as much as intelligence.»

Continue reading “OpenAI introduces Codex-Spark, greatly improving coding speed”

Nvidia standardizing on GPT-5.3-Codex internally for ~30k engineers

OpenAI just scored a big win for its coding platform. (Picture: generated)
OpenAI CEO Sam Altman has been touting the latest Codex coding model all week, praising the team and extolling how excited everyone is. Adoption is also growing, and now Nvidia is voting with their feet:

Codex is now rolling out to all engineers at the company, in close cooperation with the OpenAI team who built in «cloud-managed admin controls» and fail-safe processing.

They even helped onboard the Nvidia team, saying «It’s shocking how quickly they’ve adopted Codex», and that they move like a giant startup.

GPT-5.3-Codex was only launched last week, and is perceived as a possible profit engine, competing with Claude Code for enterprise customers.

New «Chat» model coming this week, Altman says in memo to staff

After enjoying great success with GPT-5.3 on Codex, it seems to be ChatGPT’s turn this week. (Picture: generated)
While touting a return to more than 10% monthly growth for ChatGPT and «insane» Codex growth, Sam Altman also said they are preparing to launch «an updated Chat model» this week, writes CNBC.

That would likely be GPT-5.3, debuted last week for Codex. It’s a blazingly fast and more capable successor to 5.2, and is said to be one of the first models used to make itself.

The new Codex model has been a great success, funneling a 60% growth in overall use just last week.

That model is also available to Free and Go users for a limited time, and is now being extended with perhaps reduced limits, Altman says:

— We want everyone to be able to try Codex and start building.

Read the full scoop on CNBC

Ads are now live on Free and Go tiers of ChatGPT in the USA

Ads are supposed to finance giving free users the latest tech and the most messages, OpenAI says. (Picture: OpenAI)
The ads will be clearly labeled, separated from ChatGPT responses and won’t influence what the Chatbot says. Chats will be «kept private» from advertisers, OpenAI says.

— Our goal is for ads to support broader access to more powerful ChatGPT features while maintaining the trust people place in ChatGPT for important and personal tasks, they write.

Paid tiers other than Go won’t be seeing any ads at all, and the stated reason for them is to give more queries and responses to the ad-supported tiers, while keeping them fast and responsive.

It’s possible to turn off ads and get limited messages, and under-18s won’t be getting ads. They will also be disabled for «sensitive topics» such as health, mental health and politics.

The test is only for the U.S. market as it stands, and the plan is for ads to make up a little less than half of OpenAIs income once they get off the ground, CNBC reports.

Read more: OpenAI’s announcement. Writeups on Engadget, The Verge, and Gizmodo.

OpenAI releases GPT-5.3-Codex, faster and more capable

The new coding model is 25% faster — letting it do long-running tasks in a shorter time frame.

It’s the first OpenAI model that was built with itself. They used early versions of it to debug, manage deployment and diagnose test results, and say they were impressed with its capabilities.

Continue reading “OpenAI releases GPT-5.3-Codex, faster and more capable”

OpenAI hires Head of Preparedness after very public job listing

Dylan Scandinaro’s profile picture on x.com. (Picture: screenshot)
Anthropic’s safety engineer Dylan Scandinaro has agreed to join OpenAI for the crucial role in ensuring OpenAI can keep growing while mitigating risks.

The job post almost instantly went viral in December, with CEO Sam Altman warning of biological and hacking risks — saying things were moving so fast they urgently needed someone for the «stressful job,» to be ready to «jump into the deep end pretty much immediately.»

On the hiring, Altman says he has found the best candidate for the job, and is «extremely excited» to welcome Scandinaro, who says there are great benefits ahead, but also warns of «irrecoverable harm» if not handled correctly.

— Things are about to move quite fast and we will be working with extremely powerful models soon. This will require commensurate safeguards to ensure we can continue to deliver tremendous benefits, Altman writes.

Read more: The Verge, Bloomberg.

Forbes profiles Sam Altman, with a tease of Artificial General Intelligence

Sam Altman at TechCrunch Disrupt 2019. (Picture: TechCrunch, CC BY 2.0)
AGI could be right around the corner, Altman teases in the interview, before Microsoft CEO Satya Nadella says «I don’t think we are anywhere close,» and Altman backtracks.

What is for certain is that AI is getting more capable with every new model, and Altman says:

— We are heading toward a system that will be capable of doing innovation on its own, I don’t think most of the world has internalized what that’s going to mean.

On AGI, he says they would need «a lot of medium-sized breakthroughs. I don’t think we need a big one.»

Go read the full profile on Forbes, that takes you through Altman’s early career, with OpenAI as «destiny,» what he thinks about hardware — and how he plans to exit one day.

OpenAI’s Codex suite gets a macOS app

For your discerning coding needs, you are no longer tied to the Codex web interface or terminal window, and can now vibe code on your own macOS app. Windows support is «comining soon.”

The app supports multitasking agents, creating and using skills, and automations.

Read more about it and get the app here.

To celebrate the launch, Codex is now available for Free and Go tiers, and paid plans get double the usage limits for «a limited time.»

OpenAI to retire whole host of legacy models in ChatGPT, including 4o

OpenAI is binning previously popular legacy models, used by 0.1% of users. (Picture: generated)
Brought back by popular demand after the turbulent release of GPT-5, OpenAI is now sending off the GPT‑4o, GPT‑4.1, GPT‑4.1 mini, OpenAI o4-mini and GPT-5 models on February 13th.

The reason for bringing them back was how clinical and cerebral, and less friendly GPT-5 had gotten, leading to a very public backlash.

After this, OpenAI spent a lot of time thinking about personality and customization — leading to GPT-5.2, which now has the overwhelming majority of users.

The previously very popular 4o model is now only used by 0.1% of users.

In their announcement today, OpenAI also says they are making progress towards a more creative version of GPT for adults, having rolled out age prediction earlier this month.

Read more: OpenAI’s announcement, discussion on r/ChatGPT.

OpenAI closing in on $100 billion funding round at $830B valuation

The big guns are all out for OpenAI’s latest funding round. (Picture: generated)
In what looks like one of the strongest funding rounds in history, OpenAI is getting investments from SoftBank and half of the Magnificent Seven.

SoftBank and Nvidia will be the largest investors, clocking in at $30 billion each, while Amazon will pitch in «potentially» $20 billion and Microsoft will contribute «less than» $10 billion, according to Reuters and The Information.

Apparently, Amazon’s investment could come with a caveat that OpenAI expands its cloud server rental with the company, which will likely not be a large hitch.

This will also be SoftBank’s second investment in OpenAI, after recently completing a $41 billion investment, and selling out Nvidia.

That would bring their holdings to $71 billion, which is still short of Microsoft’s reported stake of $135 billion.

Read more: Reuters, and The Information, summarized by Reuters.

OpenAI launches Prism, a research and collaboration tool for scientists

Prism is billed as a one stop shop for researchers, combining what was previously spread far and wide. (Picture: OpenAI)
Scientists are often plagued by having to use different platforms and apps for seemingly mundane things.

That’s where OpenAI’s new service comes in — as a LaTeX-native interface, it combines the paper writing in the same space as it does equations, references and «surrounding context.»

Of course there is a GPT-5.2 engine right there, so it’s easy to cross-reference and check for originality, find citations and proofread — and it’s a fairly nice research assistant, to boot.

Prism is available as of now for anyone who has a personal account with OpenAI.

Read more: OpenAI’s launch page, launch tweet, The tool itself, writeup on Gizmodo.

OpenAI says ChatGPT increasingly used in hard science

Growing rapidly; ever more people are using ChatGPT in scientific fields. (Picture: OpenAI)
In a report shared with Axios, OpenAI is touting ChatGPT’s prowess as a research assistant, saying it has «progressed past competition level performance toward mathematical discovery.»

The scales are still low — with 1.3 million users discussing «advanced hard science,» and an average of 8.3 million weekly messages on the topics.

To put this into perspective, an October, 2025, survey from OpenAI said that 0.15% of ChatGPT users engaged in conversations on self-harm and suicide, or roughly 1.2 million customers.

Continue reading “OpenAI says ChatGPT increasingly used in hard science”