ChatGPT 5.4 model: OpenAI has introduced a new artificial intelligence model called GPT-5.4, which is now available in ChatGPT, the company’s API, and its coding tool Codex. The company said the model is designed to handle complex professional tasks more efficiently, including coding, research, and document-related work.
What is ChatGPT GPT-5.4?
GPT-5.4 is the latest generation of language model developed by OpenAI. It combines improvements in reasoning, coding ability, and workflow automation into a single system. The model also integrates the coding capabilities of GPT-5.3-Codex, which previously focused on software development tasks.
In ChatGPT, the new version appears as “GPT-5.4 Thinking.” The model is designed to provide a short plan of how it will approach a problem before generating the final response. This allows users to adjust instructions while the system is working, which can help produce results that better match their needs.
OpenAI has also announced a more powerful version called GPT-5.4 Pro, which will be released for users who require higher performance for complex work.
Improvements in long tasks
One of the key updates in GPT-5.4 is its ability to handle longer and more complex tasks. The model supports up to one million tokens of context, allowing it to process large amounts of information at once. This makes it more capable of analysing long documents, research materials, or large datasets.
The system is also designed to improve deep web research, especially when answering highly specific questions. According to OpenAI, the model can maintain context more effectively during long conversations or complex reasoning tasks.
New capabilities
GPT-5.4 is also the first general-purpose OpenAI model with built-in computer-use capabilities. In tools such as Codex and the API, the model can interact with software environments and carry out tasks across different applications.
For example, it can generate code to operate computers through automation libraries like Playwright or perform actions such as mouse clicks and keyboard commands based on screenshots.
These capabilities allow AI agents to plan and execute multi-step workflows, such as editing documents, managing spreadsheets, or interacting with online tools.
Performance and results
OpenAI claims that GPT-5.4 shows improved performance across several industry benchmarks. In the GDPval test, which measures AI performance in knowledge-based professional tasks across 44 occupations, GPT-5.4 matched or exceeded human professionals in 83 percent of comparisons.
The model also achieved higher success rates in computer-interaction tests such as OSWorld-Verified and WebArena-Verified, which measure how effectively AI systems navigate software interfaces and web browsers.
Overall, GPT-5.4 is aimed at combining coding skills, reasoning ability, and tool use into a single system capable of supporting complex digital work.
