OpenAI launched its latest artificial intelligence model, GPT-5.4, this week, with the company stating that power users of its ChatGPT platform should experience immediate improvements. The release follows the recent introduction of GPT-5.3 Instant, designed for more natural, conversational interactions.
According to OpenAI, GPT-5.4 represents its “most capable and efficient frontier model for professional work.” The company simultaneously announced GPT-5.4 Thinking and GPT-5.4 Pro, targeted at users prioritizing peak performance. The modern models are rolling out gradually in ChatGPT and Codex, and via the API, requiring a paid ChatGPT subscription for access.
One key advancement is improved efficiency. OpenAI reports that GPT-5.4 is its “most token-efficient reasoning model yet,” utilizing significantly fewer tokens to solve problems compared to GPT-5.2. This translates to reduced computational costs and faster response times, according to the company.
GPT-5.4 also demonstrates enhanced capabilities in real-world knowledge work applications. OpenAI states the model achieves a state-of-the-art performance on GDPval, a benchmark assessing agents’ abilities across 44 occupations, matching or exceeding industry professionals in 83.0% of comparisons. GPT-5.2 achieved 70.9% on the same benchmark.
OpenAI has also focused on improving accuracy. The company claims GPT-5.4 is 33 percent less likely to make false claims and its statements are 18 percent less likely to contain errors when compared to GPT-5.2.
A new feature allows users to guide the model’s reasoning process in real-time. In ChatGPT, GPT-5.4 Thinking can now present an upfront plan of its thought process, enabling users to adjust the model’s course mid-response to achieve more aligned results without requiring additional prompts.
Enterprise customers can now leverage GPT-5.4 through the ChatGPT for Excel add-in. OpenAI asserts the latest models will be more adept at creating and analyzing spreadsheets.
The launch of GPT-5.4 comes less than a week after the release of GPT-5.3 Instant, which OpenAI described as aiming to make ChatGPT “less cringe and more natural.” OpenAI last upgraded its frontier model in December with GPT-5.2, responding to competition from Google’s Gemini.
Ziff Davis, the parent company of Mashable, filed a lawsuit against OpenAI in April 2025, alleging copyright infringement in the training and operation of its AI systems.
