OpenAI Launches GPT-5.4

On March 5, 2026, OpenAI announced the release of its latest frontier model, GPT-5.4. Positioned as the most capable and efficient model for professional work to date, GPT-5.4 is being rolled out in ChatGPT as "GPT-5.4 Thinking," as well as through the API and Codex. Additionally, "GPT-5.4 Pro" has been released for users who require maximum performance on complex tasks.

One of the most significant features of GPT-5.4 is its advanced reasoning and interactivity. GPT-5.4 Thinking, integrated into ChatGPT, introduces a new capability that allows it to present an upfront plan of its thinking process, enabling users to adjust the course of the response mid-stream. This allows users to arrive at a final output that is more closely aligned with their needs with fewer interactions. The model also boasts improved deep web research capabilities for highly specific queries and better context maintenance for questions that require longer thought processes. These enhancements result in higher-quality answers that are delivered faster while remaining relevant to the task at hand.

Developer-focused features have also been substantially upgraded. GPT-5.4 is the first general-purpose model with native, state-of-the-art computer-use capabilities, allowing agents to operate across applications and execute complex workflows. It supports a context of up to 1 million tokens, enabling agents to plan, execute, and verify tasks over long horizons. The introduction of "tool search" allows the model to work efficiently even when given a large number of tools. This eliminates the need to include all tool definitions in the prompt, thereby reducing token usage and increasing response speed.

In performance evaluations, GPT-5.4 has demonstrated remarkable results. On "GDPval," a benchmark that tests knowledge work across 44 occupations, GPT-5.4 matched or exceeded industry professionals in 83.0% of comparisons, a significant improvement from GPT-5.2's 70.9%. OpenAI has also focused on reducing misinformation; individual claims made by GPT-5.4 are 33% less likely to be false, and its full responses are 18% less likely to contain any errors compared to GPT-5.2. These advancements signify that GPT-5.4 enables more reliable agents, faster developer workflows, and higher-quality outputs.