Tag Archives: OpenAI

OpenAI Unveils o3 and o3-Mini Trained To “Think” Before Responding



OpenAI saved its biggest announcement for the last day of its 12-day “shipmas” event, TechCrunch reported.

On Friday, the company unveiled o3, the successor to the o1 “reasoning” model it released earlier in the year. O3 is a model family, to be more precise — as was the case with o1. There’s o3 and o3-mini, a smaller, distilled model fine-tuned for particular tasks.

OpenAI makes the remarkable claim that o3, at least in certain conditions, approached AGI — with significant caveats.

Why call the new model o3, not o2? Well, trademarks may be to blame. According to The Information, OpenAI skipped o2 to avoid a potential conflict with British telecom provider O2. Sam Altman somewhat confirmed this during a livestream this morning. 

Neither o3 nor o3-mini are widely available yet, but safety researchers can sign up for a preview for o3-mini starting today. An o3 preview will arrive sometime after; OpenAI didn’t specify when. Altman said that the plan is to launch o3-mini toward the end of January.

NBC News reported: OpenAI’s “12 days of Shipmas” which wrapped up on Friday, brought a sense of levity to end the year. The marketing blitz served as a way for the high-profile and controversial AI startup to show it can release an extensive roster of new features and tools while also having fun.

But when the calendar turns, the company faces some serious challenges. Most notably, there’s co-founder Elon Musk, who now runs rival startup xAI, and is in the midst of a heated legal battle with OpenAI CEO Sam Altman that could have a big impact on the company’s future.

The threat Musk poses to OpenAI is even more significant considering the hefty amount of influence the world’s richest person is poised to assume as part of the incoming Trump administration.

The pressure on OpenAI is tied in large part to its $157 billion valuation, achieved in the two years since the company launched its viral chatbot, ChatGPT, and kicked off the boom in generative AI.

ArsTechnica reported: Sam Altman announced its latest AI “reasoning” models, o3 and o3-mini, which build upon the o1 models launched earlier this year. The company is not releasing them yet but will make this models available for public safety testing and research access today.

The models use what OpenAI calls “private chain of thought,” where the model pauses to examine its internal dialogue and plan ahead before responding, which you might call “simulated reasoning” (SR) – a form of AI that goes beyond basic large language models (LLMs).

The o3-mini variant, also announced Friday, includes an adaptive thinking time feature, offering low, medium, and high processing speeds. The company states that higher compute settings produce better results. OpenAI reports that o3-mini outperforms its predecessor, o1, on the Codeforces benchmark.

In my opinion, it sounds like OpenAI o3 and o3-mini are likely to become the next generation of AI models.


Meta Urges California Attorney General To Stop OpenAI From Becoming For-Profit



Meta Platforms is urging California’s attorney general to block OpenAI’s planned conversion to a for-profit company, siding with Elon Musk in a battle between Silicon Valley’s most powerful artificial-intelligence players, The Wall Street Journal reported.

In a letter to Attorney General Rob Bonta dated Thursday, Meta said allowing the ChatGPT maker to become a for-profit company would set a dangerous precedent of allowing startups to enjoy the advantages of nonprofit status until they are poised to become profitable.

“OpenAI’s conduct could have seismic implications for Silicon Valley,” Meta wrote in the letter. “If OpenAI’s new business model is valid, non-profit investor would get the same for-profit upside as those who invest the conventional way in for-profit companies while also benefiting from tax write-offs bestowed by the government.”

OpenAI Chairman Bret Taylor said there will continue to be a nonprofit arm of OpenAI following a potential restructuring that will receive full value in its ownership stake of the for-profit, along with “an enhanced ability to pursue its mission” of ensuring AI benefits humanity.

Meta is one of OpenAI’s biggest competitors and has invested billions of dollars to develop its own AI technology that matches or exceeds ChatGPT. OpenAI is also closely allied with two large Meta rivals: Microsoft, which is OpenAI’s biggest investor, and Apple, which integrated ChatGPT into its own AI product.

TechCrunch reported Facebook’s parent company Meta is supporting Elon Musk’s effort to block OpenAI’s conversion from a non-profit company to a for-profit one.

As reported by The Wall Street Journal, Meta sent a letter to California Attorney General Rob Bonta arguing that allowing the shift would have “seismic implications for Silicon Valley.” And while the company asked Bonta to take “direct action,” it also said Musk and former OpenAI board member Shivon Sills “are qualified and well-positioned to represent the interests of Californians in this matter.”

“If OpenAI’s new business model is valid, non-profit investors would get the same for-profit upside as those who invest the conventional way for-profit companies while also benefiting from tax write-offs bestowed by the government,” Meta wrote.

Meta has become one of OpenAI’s big competitors in the AI market. Musk, meanwhile, was originally a co-founder at OpenAI before eventually splitting from the company, starting rival xAI, and taking legal action that includes seeking an injunction to stop its transition for a for-profit.

Engadget reported Meta sent a letter to California’s Attorney general on Thursday urging him to stop OpenAI from converting to a for-profit company, a move Meta says would be “wrong” and “could lead to a proliferation of similar start-up ventures that are notionally charitable until they are potentially profitable. 

The letter from Meta Platforms to Attorney General Rob Bonta, first reported on The Wall Street Journal, comes on the heals of an injunction filed by Elon Musk at the end of November that also asked for OpenAI’s conversion to be blocked.

In my opinion, I think Meta Platforms is trying to convince Attorney General Rob Bonta to prevent OpenAI from becoming for-profit.


Inside Elon Musk’s Quest To Beat Open AI At It’s Own Game



Elon Musk spent the past year building his artificial intelligence startup xAI at breakneck speed. Now he has to turn it into a real business, The Wall Street Journal reported.

Musk started xAI last summer in an effort to play catch-up with OpenAI, the ChatGPT developer he co-founded and left in 2018 after a power struggle. He poached talent from across the industry. He pushed contractors to build a massive new data center in a matter of months, a nearly unheard of timeframe for a project of that size.

Now, he is promising the facility in Memphis, Tenn. will help xAI deliver the world’s most powerful AI “by every metric” by December.

Most of xAI’s revenue has come from Musk’s own web of companies. xAI’s main product – its Grok chatbot – is available only to subscribers of his social network X. The startup is powering customer support features for SpaceX’s Starlink Internet service, people with knowledge of the matter said. It is also expected to help create new AI features for X’s search engine, one of the people said.

The startup has discussed deal with Tesla whereby xAI would get some Tesla revenue in exchange for providing the carmaker with access to its technology and resources.

TechCrunch reported Elon Musk’s AI company, xAI, is reportedly preparing a release a stand-alone consumer app.

The app will likely arrive after xAI closes its next funding round, which could reach $5 billion and value the company $50 billion, per the Financial Times — double its valuation six months ago. Musk is said to have given investors who backed his $44 billion Twitter acquisition 25% (or access up to 25%) of the shares in xAI to reward their loyalty.

According to the FT, some of Musk’s backers — including Fidelity, Oracle co-founder Larry Ellison, and Twitter founder Jack Dorsey — could be made whole through shares in xAI thanks to the startup’s massive rise in value.

The Verge reported Elon Musk’s xAI could soon make its next move to compete with OpenAI: launching a standalone app for its Grok chatbot.

Musk created xAI as an alternative to OpenAI, the company he helped found but later distanced himself from over ideological differences. Now, The Wall Street Journal reports, xAI is planning to launch an app as soon as December that could go head-to-head with Open AI’s ChatGPT as it races to scale up. 

Currently, users can access Grok through X, but only if they subscribe to the service. Citing unnamed sources, the Journal reports that xAI is also behind customer support features for Starlink, part of Musk’s other company, SpaceX.  X did not immediately respond to a rust for comment on the report.

In my opinion, it looks like Elon Musk is very interested in making use of xAI right now. I’m not entirely sure what he plans to do with that.


OpenAI Plans To Release Its Next Big Model By December



OpenAI plans to launch Orion, its next frontier model, by December. The Verge reported.

Unlike the release of OpenAI’s last two models, GPT-4o and o1, Orion won’t initially be released widely through ChatGPT. Instead, OpenAI is planning to grant access first to companies it works closely with in order for them to build their own products and features, according to a source familiar with the plan.

Another source tells The Verge that engineers inside Microsoft — OpenAI’s main partner for deploying AI models — are preparing to host Orion on Azure as early as November. While Orion is seen inside OpenAI as the successor to GPT-4, it’s unclear if the company will call it GPT-5 externally. As always, the release plan is subject to change and could slip. Microsoft declined to comment for this story, and OpenAI initially declined.

After CEO Sam Altman called this story “fake news,” OpenAI spokesperson Niko Felix told The Verge that the company doesn’t “have plans to release a model code-named Orion this year” but that “we do plan to release a lot of other great technology.”

TechCrunch reported OpenAI says that it doesn’t intend to release an AI model code-named Orion this year, countering recent reporting on the company’s product roadmap.

“We don’t have plan to release a model code-named Orion this year,” a spokesperson told TechCrunch via mail. “We do plan to release a lot of other great technology.”

OpenAI previously told TechCrunch that The Verge’s report wasn’t accurate, but declined to elaborate further.

Orion, a step up from OpenAI’s current flagship, GPT-4o, is reportedly trained in part on synthetic training data from o1, the company’s “reasoning” model. Open AI plans for the foreseeable future to continue developing new “GPT” models alongside reasoning models like o1, which it sees as addressing fundamentally different use cases.

Open AI’s statement leaves substantial wiggle room. It could be that the company’s next major model isn’t, in fact, Orion. Or perhaps, OpenAI will release a new model by December, but one less capable than Orion.

TechRadar reported OpenAI CEO Sam Altman has slammed a media report about the imminent release of Orion, which is effectively ChatGPT-5, with a terse tweet on X.com. He described the report as “fake news out of control,” squashing rumors of a new version of ChatGPT before December.

The report by The Verge quotes ‘sources’ as claiming Microsoft is planning to host Orion, the successor to ChatGPT-4 on its servers in November, pointing towards a release for the new LLM in time for ChatGPT’s 2nd birthday next month.

Many users took this to mean a new product launch was imminent around the time of ChatGPT’s 2nd birthday (ChatGPT was released on 30 November 2022). Altman’s latest X.com post however would seem to remove any ambiguity around the event.

In my opinion, it appears that various news sites have either gotten their news regarding Orion from sources that are unnamed. As such, we will have to wait and see what actually happens.


Apple No Longer In Talks To Join OpenAI Investment Round



Apple is no longer in talks to participate in an OpenAI funding round expected to raise as much as $6.5 billion, an 11th hour end to what would have been a rare investment by the iPhone maker in another major Silicon Valley company, The Wall Street Journal reported.

Apple recently fell out of the talks to join the round, which is slated to close next week, according to a knowledgable person.

Two other tech giants, Microsoft and Nvidia, have also been in talks to participate in the round. Microsoft is expected to invest around $1 billion, adding to the $13 billion it already has put into the company, according to people familiar with the matter.

The funding talks aren’t complicated and it is possible the participants and investment amounts could change.

OpenAI is also in the process of overhauling its corporate structure from a nonprofit into a for-profit company. That change, which was encouraged by many of the investors in the round, will be a complicated process for the startup. If it doesn’t complete the change within two years, investors in the current round will have the right to request their money back.

9To5 Mac reported Apple was first reported to be discussing an OpenAI investment last month. It was never clear, however, how much Apple might’ve invested in the company. 

Microsoft, of course, is already a major investor in OpenAI and owns a 49% share of its profits. Today’s report suggests Microsoft is set to expand its OpenAI investment with another $1 billion through this latest round.

The news of Apple’s decision to pass on investing in OpenAI comes amid more leadership and structural changes at the company. Former Apple design chief Jony Ive also recently confirmed that he is working with OpenAI to design an AI hardware product.

Apple and OpenAI are teaming up to integrate ChatGPT into iOS 18 later this year. Through this partnership, ChatGPT will handle world knowledge requests in combination with Siri.

Mashable reported Apple is reportedly no longer in talks with ChatGPT creator OpenAI for investing in its latest fundraising round.

The Washington Post reported on Friday that a “knowledgable person” reported that Apple fell out of talks to join this round. It’s a surprise because the news came at such a late hour — the talks are scheduled to wrap up this upcoming week — but should’t shock anyone too intensely.

Apple rarely invests in another Silicon Valley company, and is known for being pretty conservative in the world of direct investments and acquisitions in comparison to other tech giants.

In my opinion, Apple can do whatever it wants to, including opting-out of supporting OpenAI.


OpenAI Release o1, Its First Model With ‘Reasoning’ Abilities



OpenAI is releasing a new model called o1, the first in a planned series of “reasoning” models that have that have been trained to answer more complex questions, faster than a human can. It’s being released alongside o1-mini, a smaller, cheaper version. And yes, if you’re steeped in AI rumors: this is, in fact, the extremely hyped Strawberry model, The Verge reported.

For OpenAI, o1 represents a step toward its broader goal of human-like artificial intelligence. More practically, it does a better job at writing code and solving multistep problems than previous models. But it’s also more expensive and slower to use than GPT-4o. OpenAI is calling this release of o1 a “preview” to emphasize how nascent it is.

ChatGPT Plus and Team users get access to both o1-preview and o1-mini starting today, while Enterprise and Edu users will get access early next week. OpenAI says it plans to bring o1-mini access to all the free users of ChatGPT but hasn’t set a release date yet.

Developer access to o1 is really expensive: In the API, o1-preview is $15 per 1 million input tokens, or chunks of test parsed by the model, and $60 per 1 million output tokens. For comparison, GPT-4o costs $5 per 1 million input tokens and $15 per 1 million output tokens. 

OpenAI posted the following:  

We’re releasing OpenAI o1-mini, a cost-efficient reasoning model. O1-mini excels at STEM, especially math and coding — nearly matching the performance of OpenAIo1 on evaluation benchmarks such as AIME and Codeforces. We expect o1-mini will be a faster, cost-effective model for applications that require reasoning without broad world knowledge.

Today, we are launching o1-mini to tier 5 API users at a cost that is 80% cheaper than OpenAI o1-preview. ChatGPT Plus, Team, Enterprise, and Edu users can use o1-mini as an alternative to o1-preview, with higher rate limits and lower latency.

Wired reported  OpenAI made the last big breakthrough in artificial intelligence by increasing the size of its models to dizzying proportions, when it introduced GPT-4 last year. The company today announced a new advance that signals a shift in approach — a model that can “reason” logically through many different problems and is significantly smarter than existing AI without a major scale-up.

The new model, dubbed OpenAI o1, can solve problems that stump existing AI models, including OpenAI’s most powerful existing model, GPT-4o. Rather than summon up an answer in one step, as a large language model normally does, it reasons through the problem, effectively thinking out loud as a person might, before arriving at the right result.

The new model was a code-named Strawberry within OpenAI, and is not a successor to GPT-4o but rather a compliment to it, the company says. 

In my opinion, OpenAI o1 could be useful for those who are teaching – or learning – how o1 works. Unfortunately, it appears that accessing it can come at a very high price point.


OpenAI Launches GPT-4o Mini



OpenAI announced it will launch a new AI model, “GPT-4o mini,” the artificial intelligence startup’s latest effort to expand use of its popular chatbot, on Thursday, CNBC reported.

The company called the new release “the most capable and cost efficient small model available today,” and it plans to integrate image, video, and audio into it later.

The mini AI model is an offshoot of GPT-4o, OpenAI’s fastest and most powerful model, which it launched in May during a live-streamed event with executives. The “o” in GPT-4o stands for omni, and GPT-4o has improved audio, video and text capabilities, with the ability to handle 50 different languages at improved speed and quality, according to the company.

OpenAI posted: GPT-4o mini: advancing cost-efficient intelligence

OpenAI is committed to making intelligence as broadly accessible as possible. Today, we’re announcing GPT-4o mini, our most cost-efficient small model. We expect GPT-4o mini will significantly expand the range of applications built with AI by making intelligence much more affordable. 

GPT-4o mini scores 82% on MMLU and currently outperforms GPT-4(1) on chat preferences in LMSYS leaderboard. It is priced at 15 cents per million input tokens and 60 cents per million output tokens, an order of magnitude more affordable than previous frontier models and more than 60% cheaper than GPT-3.5 Turbo.

GPT-4o mini enables a broad range of tasks with its los cost and latency, such as applications that chain or parallelize multiple model calls (e.g., calling multiple APIs), pass a large volume of context to the model (e.g., full code base or conversation history), or interact with customers through fast, real-time text responses (e.g., customer support chatbots).

Today, GPT-4o mini supports text and vision in the API, with support for text, image, video and audio inputs and outputs coming in the future. The model has a context window of 128K tokens, supports up to 16K output tokens per request, and has knowledge up to October 2023. Thanks to the improved tokenizer shared with GPT-41 , handling non-English text is now even more cost effective…

TechCrunch  reported OpenAI introduced GPT-4o mini on Thursday, its latest small AI model. The company says GPT-4o mini, which is cheaper and faster than OpenAI’s current cutting edge AI models, is being released for developers, as well as through the ChatGPT web and mobile app for consumers starting today. Enterprise users will gain access next week.

The company said GPT-4o mini outperforms industry leading small AI models on reasoning tasks involving text and vision. As small AI models improve, they are becoming more popular for developers due to their speed and cost efficiencies compared to larger models such as GPT-4 Omni or Claude 3.5 Sonnet. They’re a useful option for high-volume simple tasks that developers might repeatedly call on an AI model to preform.

In my opinion, it might be better to use a smaller version of GPT-4o than to make larger ones that require things that humans need. More specifically, I have concerns about how much water gets used by AI systems.