OpenAI on Thursday announced GPT-5, the foundational model for the next generation of ChatGPT and a “significant leap in intelligence,” according to the company.
GPT-5 will be released in two versions, a “pro” model which will only be accessible to paid subscribers, and the basic GPT-5, which will actually be available to everyone, even those on ChatGPT’s free plan. Plus subscribers will get more usage, OpenAI said.
Interestingly, OpenAI won’t explicitly make which model ChatGPT is using available to ChatGPT users. An interesting feature of the new GPT-5 is a “real-time router” which will assign a model based on the query the user asks. If you use the terms “think hard about this,” ChatGPT might see it as a signal that a more complex model needs to be used. One “efficient” model may be used for most questions, with a deeper, thinking model available if needed. Apparently, there’s a “mini” version as well that will kick in if a user exceeds the usage rates.
“We’ve made significant advances in reducing hallucinations, improving instruction following, and minimizing sycophancy, while leveling up GPT‑5’s performance in three of ChatGPT’s most common uses: writing, coding, and health,” OpenAI said in a blog post.
One area of expertise is coding, where ChatGPT and GPT-5 can now create entire websites and games with just a single prompt, OpenAI said, while surpassing many of its previous levels of expertise in health, too. Our sister site, TechCrunch, has a lot more on GPT-5.
