OpenAI, the US-based AI company behind ChatGPT, is expected to unveil its next-generation language model, GPT-5, later today. The company shared a teaser on X (formerly Twitter) on Wednesday that read: “LIVE5TREAM THURSDAY 10AM PT,” with the letter ‘S’ replaced by a ‘5’ — strongly hinting at the debut of GPT-5. The livestream is set for 10:30 PM IST.

The announcement follows a busy month for OpenAI, during which it introduced mental health-focused upgrades to its platform and released two new open-weight AI models capable of running locally on personal computers. Now, attention has shifted to the potential reveal of GPT-5. Rumors of a launch in early August have been circulating for weeks, and this teaser seems to confirm it may happen tonight. There are indications that Microsoft has already begun testing GPT-5 within its Copilot assistant. According to a report from TestingCatalog, GPT-5 features are being trialed inside Copilot, suggesting that Microsoft might integrate the model shortly after its official release.

According to The Verge, GPT-5 is expected to move beyond the standard single-model structure. Instead, it will likely adopt a modular design that integrates multiple specialized models to handle different tasks more efficiently. OpenAI CEO Sam Altman has previously mentioned that GPT-5 would build upon the o3 model and incorporate other advanced technologies to create a more cohesive and powerful AI ecosystem. In a recent reply to a user on X who asked whether GPT can now recommend TV shows, Altman simply said: “Turns out yes!”

Eventually, OpenAI aims to merge its GPT and o-series models into a unified architecture to support a broader range of features and tools. While GPT-5 will likely offer basic functionality to free ChatGPT users, those subscribed to Plus and Premium tiers are expected to gain access to its most advanced capabilities, including higher reasoning and performance levels.