OpenAI cancels its o3 AI model in favor of a ‘unified’ next-gen release

OpenAI has effectively canceled release O3On behalf of the CEO Sam Altman calling a “simplified” product offer, the company’s next Major AI model was ready.

A Post at x on WednesdayUltman said that in the coming months Openi would publish a model called GPT -5 that integrates “abundant amounts [OpenAI’s] Technology, “on the AI-driven chattbot platform and 3 Chatzipt And API. As a result of that roadmap’s decision, OpenAI does not plan to launch 3 as 3 Stand-1 models.

The company originally said in December that it was aimed at publishing the O3 early this year. Just a few weeks ago, Openai’s Chief Product Officer Kevin Well, In an interview said O 3 was on the track for a “February-March” launch.

“We want to do better to share our intended roadmaps and do better to work better to facilitate our product offers,” Writes in the post. “We want II to ‘just work’ for you; We understand how complicated our models and product offers have been. We hate model picker [in ChatGPT] As much as you do and Magic Unified Intelliged Want to return. “

Ultman also announced that the model is usually available, once subject to “abuse thresholds”, on OpenAI “Standard Intelligence Setting” in OpenAI “Standard Intelligence Setting”. (Ultman refuses to provide more details about the thresholds of this abuse and the threshold of this abuse)) ChatzPT Plus customers will be able to “run GPT-5 at the higher intelligence level,” Ultman says when the ChatzPT Probles will be a high level of intelligence. “A capable of running GPT -5.

“[GPT-5] Will include voice, canvas, search, deep research and more, ”he added, OpenAI mentions the various features launched in the Chatzipi for the past few months. “[A] To integrate the top goals for us [our] Models by creating models that can use all our tools, know when to think for a long time, and are usually useful for very broad jobs ”

Before introducing GPT-5, OpenAI plans to publish its GPT-5.1 model, code-called “Orion”, according to X-Altman’s post, Ultman says that it is the company’s latest “chain-off-off-off-off. -Off will be “-chal -thought model. “Unlike other rational models of O and 3 and OpenAE, non-chain-thought models are less reliable in domains such as mathematics and physics.

Looks like Opina is fully embracing the rational model trend. It is arguably cockestart with its first rational model, O 1By the end of last year. Rational models effectively testify their truth, which Usually helps them avoid some damage that models travelThe This fact-checking process may be somewhat delayed-addict models take a little longer to reach the solutions, usually a few minutes more time. However, they both are more reliable and able to be able.

The Chinese AI Lab DIPSEC has recently attracted the world’s attention to its R1 model, which matches a number of criteria and 1. In contrast to O 1, the R1 is a “open” model under a permit license, which means downloading and developers can be used as suitable.

In recent social media posts, Ultman Admission That DIPSEC AI reduced the technical leadership of the Openai in AI and D That OpenAI will “pull some release” to compete better.

GPT -4.5, or OreononIt is said that several performance-related challenges and technical disasters have suffered. Blomberg, InformationAnd Wall Street Journal Indeedly revealed that Orion showed less improvement than his predecessor, GPT -4OGPT -1 is higher than GPT -1.