Welcome to Neural. AI is evolving rapidly. We’re here to help you stay informed. Just earlier this week, Elon Musk initiated a move to acquire OpenAI, which led to Sam Altman informally declining the offer, while cheekily suggesting a bid for Twitter instead. Although this was quite amusing, the AI landscape seems to be settling down—WAIT, WHAT?! OpenAI has just revealed ChatGPT-5, set to launch this year?
Sam Altman, the head of OpenAI, has effectively regained control of the news with a detailed overview of the future of ChatGPT. He shared what he called an OpenAI roadmap update concerning both the GPT-4.5 and GPT-5 AI models.
Altman begins by addressing the current intricacies within the ChatGPT experience.
We aim to improve our communication regarding our roadmap and significantly simplify our product offerings. Our goal is for AI to “just work” for users; we recognize how convoluted our model and product selection has become.
A rapid succession of product releases has contributed to the complexity within ChatGPT. Currently, for ChatGPT Plus subscribers, the app features the following model options and descriptions:
- GPT-4o: Suitable for most inquiries
- GPT-4o with scheduled tasks: Allows follow-up requests
- o1: Employs advanced reasoning
- o3-mini: Quick at advanced reasoning
- o3-mini-high: Excellent for coding and logical tasks
- GPT-4o mini: Faster responses for most questions
- GPT-4: Legacy version
The challenge for users lies in identifying the appropriate model to begin with: a large language GPT model or a reasoning o-series model. OpenAI’s challenge is determining which model will yield the necessary response. This is something OpenAI is currently tackling.
We share your frustration with the model selection process and aspire to return to a seamless, unified intelligence. Our next release will be GPT-4.5, the model we internally referred to as Orion, marking our last non-chain-of-thought model. After this, a primary objective will be to integrate o-series and GPT-series models, creating systems capable of utilizing all our tools, discerning when deeper reasoning is required, and generally being useful across a broad spectrum of tasks.
In his post, Altman clarified that the previously hinted Orion model will be GPT-4.5, arriving within weeks. While it will not feature the DeepSeek-like chain-of-thought mechanism available in o3, ChatGPT-5 and future versions will incorporate this to illustrate the model’s problem-solving process.
According to Altman, ChatGPT-5 will serve as a cohesive model that connects LLM and reasoning models, with free chat functionalities included. He added that ChatGPT-5 is anticipated in a few months.
In both ChatGPT and our API, GPT-5 will be released as a system that integrates various technologies, including o3. We will phase out o3 as an independent model. Users on the free tier of ChatGPT will receive unlimited chat access to GPT-5 at standard intelligence levels, subject to usage limitations.
Finally, Altman emphasized that GPT-5 will provide a higher intelligence level for paid users, likely indicating more demanding capabilities.
Plus subscribers will have access to GPT-5 with enhanced intelligence levels, while Pro subscribers will enjoy an even higher intelligence setting. These models will feature voice capabilities, canvas tools, search functions, deep research, and much more.
I must admit, my one critique of Altman’s note is that it wasn’t composed entirely in lowercase letters. Seems a bit off to me.
In a prior update, Altman indicated that OpenAI plans to allow up to 10 deep research queries monthly for the $20/month tier, while free users will start with two research queries. Currently, deep research is tied to OpenAI’s $200/month ChatGPT Pro plan.
Nothing capitalized for that one.
Stay tuned for more updates regarding the latest AI advancements in the forthcoming edition of Neural—exclusively on DMN! You can read the previous issue here.
