OpenAI releases GPT-5.4 and GPT-5.4Pro models

LightningPacketLoss · 2026-03-08T02:44:30+00:00

OpenAI launched GPT-5.4 and its Pro version on March 5th, enhancing thinking modes and deep web search capabilities, supporting collaboration with various tools and software environments. The model features up to 1 million tokens of context processing and complex workflow management capabilities.

LightningPacketLoss

2026-03-08 02:44:30

Abstract generation in progress

On March 5th, OpenAI officially launched GPT-5.4, offering a “Thinking” mode, and released it on ChatGPT, API, and Codex platforms. At the same time, a GPT-5.4 Pro version was introduced for ChatGPT and API platforms.

OpenAI states that GPT-5.4 not only inherits the programming capabilities of GPT-5.3-Codex but also optimizes the model’s collaborative performance across various tools, software environments, and professional tasks. In ChatGPT, GPT-5.4 Thinking can now display its thought process in advance, allowing users to adjust the direction in real-time during system processing. GPT-5.4 Thinking also enhances deep web search capabilities. OpenAI indicates that GPT-5.4 has native cutting-edge computer operation abilities, enabling agents to control computers across applications and execute complex workflows. The model supports context processing of up to 1 million tokens. (Interface)

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.

2 Likes