OpenAI releases GPT-5.4 and GPT-5.4 Pro models

K-LinePoet · 2026-04-06T19:31:03+00:00

OpenAI has released GPT-5.4, introducing a "Thinking" mode and optimizing GPT-5.4 Pro, enhancing deep web research capabilities and context retention, allowing users to interrupt and adjust responses. The new features are now available on Android and web platforms, with the iOS version coming soon. GPT-5.4 also has native computer usage capabilities, supporting efficient reasoning for complex tasks.

K-LinePoet

2026-04-06 19:31:03

Abstract generation in progress

OpenAI officially releases GPT-5.4, offering a “Thinking” mode, and rolls it out across the ChatGPT, API, and Codex platforms. It also launches GPT-5.4 Pro, optimized for complex tasks. OpenAI says that in ChatGPT, GPT-5.4’s thinking has improved in deep-web research, with better context retention during long thinking sessions—now users can interrupt it while the model is responding, add instructions, or change direction. The guidance feature is already live this week on Android and the web; the iOS version is about to be released. OpenAI developers say that GPT-5.4 has native computer-use capabilities. Codex and API support up to 1 million context tokens, enabling best-agent coding for complex tasks, providing scalable tool search across a larger ecosystem, and delivering more efficient reasoning for long, tool-intensive workflows.

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.