On Thursday, The Verge disclosed that OpenAI’s ChatGPT 5 is set to launch in August. Although a summer rollout was anticipated, lengthy model development can often lead to delays. This news is encouraging for enthusiasts of the widely-used AI chatbot, despite the lack of extensive official information.

Nevertheless, there is no shortage of unofficial insights, rumors, and leaks.

What can we anticipate with ChatGPT 5?

ChatGPT 5, Unpacked

If you use ChatGPT occasionally, the majority of your inquiries are probably addressed by GPT-4o, the standard version that OpenAI labels as “excellent for most tasks.” OpenAI is continually enhancing its offerings, having quickly released new AI tools in 2025. Whether this rapid development can proceed given Meta’s recent recruitment of AI talent is still uncertain.

This year, OpenAI introduced a celebrated image generation tool, followed by two new reasoning models, o3 and o4-mini, in April. They also launched the Operator AI agent, an educational collaboration with Microsoft and educators, along with a more advanced reasoning model, o3-pro. Beyond ChatGPT-5, OpenAI revealed an AI wearable in partnership with iPhone designer Jony Ive and is reportedly developing an AI browser, among other initiatives.

Significant enhancements to the ChatGPT usage experience are expected with ChatGPT-5.

What Do We Know About GPT-5?

Officially, very little is known apart from the fact that GPT-5 is currently in testing. On July 23, OpenAI CEO Sam Altman spoke about GPT-5 with podcaster Theo Von, stating, “5.0 I believe will be impressive.”

Altman recounted his experience testing the new model: “This morning I was trying out our new model, and I received a question via email that I didn’t quite grasp. I entered it into the model, this GPT5, and it responded flawlessly. I honestly leaned back in my chair, having an ‘oh wow’ moment. I quickly moved on to my next task, but it felt… I felt somewhat inadequate compared to the AI accomplishing something I thought I should handle. It was challenging, but the AI made it seem effortless. Yeah, it was a strange sensation.”

AI leaders often offer similar stories when launching new versions.

We anticipate GPT-5 will deliver enhanced agentic functionalities.

ChatGPT-5 May Merge Various Models and Tools

At present, ChatGPT users can switch between models, ranging from GPT-4 to advanced reasoning systems. OpenAI also provides tools like Sora for video creation and Operator, a web-based AI assistant.

Tech Radar indicates early testers have suggested that GPT-5 will unify these tools into a single interface, enabling users to create images, conduct thorough research, or code seamlessly.

Improved Coding Skills

Earlier this month, Altman shared a post on X, which many believe pertains to GPT-5.

Major model launches typically introduce gradual enhancements in fundamental skills such as image generation and coding, which is anticipated with OpenAI’s latest release.

GPT is “Coming Soon,” Promises OpenAI Leaders

Altman is not the only one evaluating GPT-5. OpenAI research scientist Alexander Wei, in the context of OpenAI’s recent math Olympiad victory, stated, “we are unveiling GPT-5 shortly.”

Tech Radar also cites independent researchers who discovered a mention of the model in “OpenAI’s internal BioSec Benchmark repository, indicating that the model is being tested in crucial fields such as biosecurity.”

Expanded Context Windows

With every new model, OpenAI broadens ChatGPT’s context window, referring to the amount of information a model can “recall” when answering a query. Developers and users are consistently asking for larger context windows, which is anticipated to be a feature in GPT-5.

OpenAI has effectively limited leaks about GPT-5. Aside from a few ambiguous hints regarding the model’s impending launch, OpenAI has kept a firm grip on messaging regarding the forthcoming model.