OpenAI — the company whose chatbot, ChatGPT, sparked the recent global interest in and usage of generative artificial intelligence — has filed a trademark application for “GPT-5” with the United States Patent and Trademark Office (USPTO).

Underpinning ChatGPT is the GPT itself — the generative pre-trained transformer, a type of large language model created by OpenAI. The most recent and publicly available version is GPT-4.

An application for the GPT-5 mark was filed by the AI company in the latter half of last July. The filing highlights that it’s in relation to “computer software for using language models.”

The filing for GPT-5 also details natural language processing, understanding, analysis, and generation, and the developing, running, and analysing of algorithms that are able to learn to analyse, classify, and take actions in response to exposure to data, among other expected features.

One particularly interesting mention in the filing is the artificial production of human speech and text. While this was also specified in the trademark application for GPT-4, it indicates that ChatGPT may have verbal capacity in the future.

Despite OpenAI’s most recent USPTO filing, it doesn’t confirm whether notable internal work is yet being done or not on GPT-5, or if it even exists yet.

Recommended

• Get Familiar With OpenAI’s Latest GPT-4
• OpenAI Chooses UK for its First International Office
• Microsoft Invests $1 Billion in OpenAI

In April, The Verge reported that Sam Altman, the AI company’s chief executive and co-founder, said at an MIT event that OpenAI wasn’t training GPT-5 and also wouldn’t be for some time. This was stated just a month after the release of GPT-4, which came out on March 14.

The GPT-5 application has been filed under a “1B” basis though, meaning that the mark hasn’t yet been used in commerce but the applicant has an intent to use the mark in commerce. Companies often file trademarks or patents for concepts that haven’t yet gotten to the development stage to stay a step ahead of competitors while also protecting their intellectual property.

Ultimately, due to the lack of official information from OpenAI, it’s unclear what stage the AI company is currently at with GPT-5, and what both the tech world and world at large could eventually expect from the model.

Relatedly, last month, amid reports of GPT-4 users complaining about degraded performance — from weakened logic to instructions not being properly followed — OpenAI said in a blog post update that “While the majority of metrics have improved, there may be some tasks where the performance gets worse.”