Previously, the company had set its sights on o3 as the next big AI model, but the CEO has simplified product offerings. In a post on X (formerly Twitter) on Wednesday, Altman shared that in the coming months, OpenAI will release a model called GPT-5, which will fold most of OpenAI’s o3 technology into its ChatGPT-based AI-powered chatbot platform and API. Thus, this roadmap decision means OpenAI will no longer launch o3 as a stand-alone model. This ended a long course of development, with the launch of o3 planned for much earlier this year and OpenAI’s Chief Product Officer, Kevin Weil, stating in a few weeks ago interview that o3 was really ready for a “February-March” launch.
Formerly it was established as its next promising AI model by the company o3. But the decision was taken by CEO Sam Altman to have simpler product offerings. In a post on X- ‘the Twitter’ on Wednesday, Altman further revealed that OpenAI will release into the wild in the forthcoming months an edition of GPT named GPT-5. It is said to incorporate most of OpenAI’s o3 technology into ChatGPT’s AI-anointed chatbot platform and API. Now according to this Road-map decision, the o3 model will not be launched as an independent model by the OpenAI. Originally, the o3 was slated to be unveiled much earlier this year, with OpenAI Chief Product Officer Kevin Weil having said a few weeks ago in an interview that o3 was “February-March”-ready.
Previously, the company had targeted o3 to be the next big model on the AI front. However, the CEO has simplified things in product offerings. On Wednesday, in a post on X (formerly known as Twitter), Altman tweeted that sometime in the next months, OpenAI would come up with a model called GPT-5 that would fold most of OpenAI’s o3 technology into its AI-powered chatbot platform, ChatGPT, and API. Therefore, as a result of this roadmap decision, OpenAI is no longer going to launch o3 as an independent model. It was originally meant for a much earlier unveiling within the year, with OpenAI Chief Product Officer Kevin Weil stating about two weeks ago in an interview that o3 was ready for a “February-March” launch.
Previously, the company established o3 as its next attempt to achieve success as a prominent AI model. But the decision was made by CEO Sam Altman to simplify the product offerings. In a post on X (formerly Twitter) on Wednesday, Altman further indicated that OpenAI would release into the wild in coming months, a version of GPT called GPT-Five as it still incorporates most of OpenAI’s o3 technology into the AI-anointed chatbot platform and API of ChatGPT. Hence according to this road-map decision, o3 will never be launched as a stand-alone model by the OpenAI. Initially, o3 was meant to be unveiled much earlier this year, with OpenAI Chief Product Officer Kevin Weil stating just a few weeks ago in an interview that o3 was ready for a “February-March” launch.
In his post, Altman explained, “We want to do a better job of sharing our intended roadmap and simplify our product offering even further. We want AI to ‘just work’ for you; we understand how complex our models and product offerings have become. We hate the model picker in ChatGPT just as much as you do, and we want to return to the magical integrated intelligence.”
Altman also announced that once OpenAI’s models are generally available, they plan to offer unlimited chat access to GPT-5 at a “standard intelligence setting,” subject to an “abuse threshold” (Altman declined to provide more details about this setting or the abuse threshold). He mentioned that ChatGPT Plus subscribers would be able to run GPT-5 at a “higher level of intelligence,” while ChatGPT Pro subscribers could run GPT-5 at an “even higher level of intelligence.” “GPT-5 will include voice, canvas, search, deep research, and much more,” Altman added, referring to several features OpenAI has recently launched in ChatGPT.
“Our top goal is to integrate our models by building systems that can use all of our tools, know when to think deeply or not, and generally be useful for a wide range of tasks,” he said.
Before the rollout of GPT-5, OpenAI is planning to release a model called GPT-4.5, codenamed “Orion,” in the next few weeks. According to Altman, this will be the company’s last “non-chain-of-thought model.” Unlike o3 and OpenAI’s other “reasoning” models, non-chain-of-thought models are less reliable in areas like mathematics and physics. It appears OpenAI is fully embracing the reasoning model trend, which it began with its first reasoning model, o1, at the end of last year. Reasoning models effectively fact-check themselves, which helps avoid some of the issues that typically trouble models. This fact-checking process takes a bit longer—reasoning models take more time to reach a solution, usually from seconds to minutes. But they are more reliable and capable.
Chinese AI lab DeepSeek recently gained some reputation with its R1 model, which matches o1 over several benchmarks. Unlike o1, R1 is an “open” model, released under a permissive license for everyone to download and do what they want. In a recent post on social media, Altman accepted that DeepSeek had decreased OpenAI’s technical edge in AI and that OpenAI was going to “pull back some releases” to enhance competition.
Meanwhile, Orion, or GPT-4.5, has been suffering from performance-related issues and technical failures for a while now. Independent reports by Bloomberg, The Information, and The Wall Street Journal suggest that Orion has improved less over its predecessor GPT-4.
