OpenAI has announced its plans for the upcoming GPT-4.5 and GPT-5 AI models, as detailed by CEO Sam Altman. The roadmap, shared on social media, states that GPT-4.5, internally known as Orion, will be the last model not utilizing a chain-of-thought approach. Altman emphasized the company’s goal to simplify its product lineup and aims to unify the various series of models to enhance their functionality across a broader range of tasks. When GPT-5 launches, users of the free version of ChatGPT will have unlimited access at a standard intelligence setting, while Plus and Pro subscribers will enjoy higher levels of intelligence. Specific release dates were not disclosed. Altman provided a vague estimate of weeks to months.
OpenAI announced a significant update to its Model Specification, expanding it from around ten pages to sixty-three pages. This new version outlines how its artificial intelligence models should engage with controversial topics while adhering to customizability, transparency, and intellectual freedom principles. The update also reflects ongoing debates in AI ethics, particularly surrounding sensitive questions like misinformation and self-harm. Notably, OpenAI aims to address the issue of “AI sycophancy,” encouraging models to provide constructive feedback rather than simply agreeing with users. OpenAI is soliciting public input on these guidelines, which will be released under a Creative Commons Zero license, allowing others in the AI community to modify and build upon them freely.
Apple has announced it will incorporate Alibaba’s artificial intelligence technology into its iPhones, highlighting a renewed confidence in the capabilities of the Chinese tech giant. Alibaba’s co-founder, Joseph Tsai, revealed this decision at the World Government Summit in Dubai, noting that Apple carefully evaluated several companies in China before selecting Alibaba. The Information reported that Apple is continuing its collaboration with Baidu to enhance artificial intelligence features for iPhone users in China, even as it partners with Alibaba Group.
Why do we care?
OpenAI’s plans for GPT-4.5 and GPT-5 reveal a shift toward product unification, but they also raise questions about accessibility and model differentiation. Model differentiation quickly goes away as these models are commoditized. And DeepSeek accelerated that process.
Note Apple looks to deliver the same features regardless of the model. Sure, the headline reason Apple is doing this is due to Chinese law. Just use that concept in the US. Perhaps Apple switches from OpenAI to DeepSeek, or something open source to save money. As long as the customer experience is the same…. And now you see the commodity play.
This drives the value into the application layer, which is suitable for customers, and the service providers serving them.