GPT-5.4 is a frontier conversational model from OpenAI that merges the GPT and Codex lines into a single system, positioning it as a unified default for both general-purpose reasoning and software engineering workloads. It is aimed at professional users who need high-context analysis, multi-step coding workflows, and instruction following across diverse tasks. Built as a successor to GPT-5.3 Instant, the model focuses on longer-running, agentic work and is designed to reduce the number of iterations needed to complete complex tasks while improving token efficiency.

GPT-5.4 is described as delivering improved performance in coding, document understanding, tool use, and instruction following, making it suitable for production-quality code generation, synthesis across multiple sources, and complex multi-step workflows. A standout feature is native computer use, which allows the model to interact with desktop and application interfaces through the API and Codex, alongside tool search for retrieving relevant tools at runtime. The very large context window supports high-context reasoning and multimodal analysis without the need to chunk large documents, which is particularly useful for long coding sessions, deep research, and enterprise workflows that previously required separate models.