OpenAI GPT-5 reportedly leaked ahead of imminent launch, variants and features tipped

OpenAI’s next-generation artificial intelligence model, GPT-5, appears to have been leaked ahead of its official unveiling. According to reports, details about multiple GPT-5 variants were published in a GitHub blog post that has since been deleted. However, screenshots circulated on social media suggest the post briefly revealed information about at least five different model versions, as well as a potential redesign for the ChatGPT interface.

The post, first flagged by TestingCatalog, reportedly listed the standard GPT-5 model along with GPT-5 Mini, GPT-5 Nano, GPT-5 Chat, and a Pro version. While OpenAI has yet to confirm the existence of these models, the leak suggests a broader product strategy, possibly aimed at catering to different levels of usage and deployment environments.

The GPT-5 Mini and Nano variants are believed to be lighter models, though technical specifications remain unconfirmed. Mini may be optimised for quicker response generation with lower latency, while Nano could be designed for use in edge computing or on-device operations, suitable for localised processing without a cloud dependency.

GPT-5 Chat and ‘Thinking’ Mode

Another notable inclusion in the leak was GPT-5 Chat, reportedly tailored for enterprise-level conversations and GitHub Copilot. Descriptions from the now-deleted post indicate it is intended to deliver more advanced, multimodal, and context-aware interactions. According to information sourced byThe Verge, GPT-5 Chat is designed for use in business applications that require sustained, nuanced dialogue.

The leak also referenced a model or feature called GPT-5 Thinking. Though details are sparse, it is speculated this may relate to an API function allowing developers to control the model’s “thinking budget”, possibly a way to allocate more computational effort to complex queries.

Can users get tiered access across ChatGPT subscriptions?

The rollout strategy, as suggested by the leak, will reportedly give users on different ChatGPT plans varying levels of access to GPT-5. Free-tier users are expected to receive limited access to the base GPT-5 model, while those subscribed to ChatGPT Plus may gain features related to advanced reasoning and increased usage limits.

Pro-tier subscribers could receive access to a new GPT-5 Pro variant, which a separate leak claims is designed to deliver “research-grade intelligence.” This particular claim was shared by a user named @scaling01 on X, who reportedly found relevant descriptions within ChatGPT’s internal codebase.

Additionally, Team plan subscribers are said to be in line for access to both GPT-5 with enhanced reasoning and the GPT-5 Pro model.

ChatGPT interface redesign on the cards?

Alongside the model rollout, a redesign of the ChatGPT user interface is also anticipated. The update is expected to feature a simplified prompt input area, a unified control button for modes and actions, and new background colour themes to enhance usability. While no images of the redesign have surfaced, TestingCatalog reports that the new interface is part of the wider GPT-5 deployment.

OpenAI has not issued an official statement confirming the leak or the details therein. However, the information aligns with growing speculation about the imminent release of GPT-5, widely expected to be announced soon.