Earlier this week, OpenAI released GPT-5.3 Instant, promising to make ChatGPT less cringe and more natural when using its most popular model. Now OpenAI is back with GPT-5.4 Thinking and Pro, which is an overall upgrade to its most capable frontier model.

OpenAI last upgraded this model in December with its GPT-5.2 “code red” response to competition from Google’s Gemini.

As for what’s new today, OpenAI highlights in these areas:

• Coding, document understanding, tool use, and instruction following
• Image perception and multimodal tasks
• Long-running task execution and multi-step agent workflows
• Token efficiency and end-to-end performance on tool-heavy workloads
• Agentic web search and multi-source synthesis, especially for hard-to-locate information
• Document-heavy and spreadsheet-heavy business workflows in customer service, analytics, and finance

GPT-5.4 touts a 1M token context window now as well:

GPT-5.4 supports up to a 1M token context window, making it easier to analyze entire codebases, long document collections, or extended agent trajectories in a single request.

OpenAI also says GPT-5.4 is its first “mainline model” with built-in computer use:

GPT-5.4 is the first mainline model with built-in computer-use capabilities, enabling agents to interact directly with software to complete, verify, and fix tasks in a build-run-verify-fix loop.

It’s also OpenAI’s first mainline model “trained to support compaction, enabling longer agent trajectories while preserving key context,” the company says.

OpenAI says GPT-5.4 is “rolling out gradually” today in ChatGPT and Codex. GPT-5.4 Thinking is available for Plus, Team, and Pro subscribers and will replace GPT-5.2 Thinking, which is going away in three months. Find out more from OpenAI’s announcement .