OpenAI has launched GPT-5.4, a new frontier model designed for professional workloads. It integrates advanced reasoning, coding, and agent-based workflows into a single system.
The model is rolling out across ChatGPT, the API, and Codex. A higher-performance GPT-5.4 Pro version is also available for complex tasks requiring maximum compute and accuracy.
GPT-5.4 enhances coding capabilities and improves performance across software environments and professional tasks like spreadsheets, presentations, and document creation. It can outline reasoning plans mid-process, allowing users to guide responses.
Native computer use capabilities enable AI agents to interact with operating systems, websites, and applications through various inputs. Developers can use GPT-5.4 to automate multi-step workflows.
The model supports context windows up to one million tokens and introduces tool search, allowing AI to dynamically locate and use external tools efficiently. Benchmarks show GPT-5.4 matching or exceeding industry professionals in 83% of comparisons on professional knowledge work, a significant improvement.
GPT-5.4 is also more efficient, requiring fewer tokens for reasoning tasks, which improves response speed and reduces costs for developers.
GPT-5.4 Thinking is available to ChatGPT Plus, Team, and Pro users. GPT-5.4 Pro is for Pro and Enterprise plans, and developers can access both models via the API.