The US-based AI company said its GPT-5.4 model integrates the latest advances in reasoning, coding, and workflow, building on capabilities from GPT-5.3-Codex.
OpenAI also introduced GPT-5.4 Pro, which it calls a more powerful version for handling complex tasks.
According to the company, GPT-5.4 improves performance in spreadsheets, presentations, and documents, enabling professionals to complete their work more efficiently. The model also supports greater context processing and better web search integration.
The GPT-5.4 Thinking version in ChatGPT can expose its reasoning flow before providing a final answer, allowing users to adjust instructions during the response process. The model is also capable of recognizing interface elements from screenshots, allowing it to perform mouse and keyboard operations across a variety of software.
In benchmark tests, GPT-5.4 outperformed previous models. It matched or surpassed industry professionals in 83 percent of GDPval comparisons and scored higher than its predecessor in coding ability tests, OpenAI said.
The model also has better factual accuracy and is less likely to generate misinformation than GPT-5.2.
GPT-5.4 is available to ChatGPT Plus, Team, and Pro users, as well as to developers via API. OpenAI plans to phase out GPT-5.2 Thinking in the coming months. While the cost per token is higher, the improved token efficiency could lower overall costs in some scenarios, the company said.
