GPT-5.4 Debuts with Native Computer Use, 1M Token Context
OpenAI announced GPT-5.4 on March 5, 2026, as its new flagship model optimized for professional work. The release brings GPT-5.4 Thinking to ChatGPT for Plus, Team and Pro users, while making the base GPT-5.4 and a higher-performance GPT-5.4 Pro available through the API and Codex. OpenAI positioned the model as its first general-purpose system with native computer-use capabilities, combining advances in reasoning, coding, tool use and agentic workflows.
The model supports a context window of up to 1 million tokens, enabling agents to handle extended, multi-step tasks across documents, spreadsheets, presentations and software environments. According to OpenAI's announcement, GPT-5.4 is more token-efficient than GPT-5.2 and better suited for professional workflows that require sustained interaction with complex files and development tools.
In coding benchmarks, OpenAI reported that GPT-5.4 matches or exceeds the performance of GPT-5.3-Codex on SWE-Bench Pro while delivering improved latency. The company highlighted stronger frontend generation capabilities and new Playwright-based interactive debugging workflows that allow the model to engage directly with browser-based testing environments. The release also introduces an enhanced "tool search" feature designed to work with larger tool ecosystems and improved browser and web search performance for agentic applications.
Technical Capabilities and Availability
GPT-5.4 is now accessible in the API under the model name gpt-5.4, with the more powerful variant available as gpt-5.4-pro. In ChatGPT, GPT-5.4 Thinking is rolling out to paid subscribers and will replace GPT-5.2 Thinking. Users who prefer the prior version can still access GPT-5.2 Thinking in the Legacy Models section until June 5, 2026.
The native computer-use features mark a significant step forward, allowing the model to interact more directly with desktop and web interfaces without relying solely on external plugins or brittle tool-calling layers. OpenAI emphasized that these capabilities were built from the ground up to support long-running professional tasks that span multiple applications and large documents.
Impact on Developers and Enterprise Users
For developers and technical teams, the combination of improved coding performance, lower latency and interactive debugging tools could accelerate software development cycles, particularly for frontend work and end-to-end testing workflows. The 1 million token context window is expected to benefit knowledge workers who need to analyze or generate content based on entire codebases, large financial models or comprehensive research documents.
Enterprise customers using OpenAI's API for agentic systems stand to gain from the expanded tool search and more reliable web browsing capabilities, potentially reducing the engineering effort required to maintain robust autonomous agents.
What's Next
OpenAI has not yet disclosed a specific timeline for further GPT-5 series updates or broader availability of the computer-use features to all user tiers. The three-month legacy window for GPT-5.2 Thinking provides organizations time to migrate workflows to the new model.
The release continues OpenAI's pattern of rapid iteration on its GPT-5 family, following previous updates that focused on reasoning and specialized coding variants. Industry observers will be watching to see how competitors respond with their own large context and computer-use offerings in the coming months.