OpenAI has introduced GPT-5.4 and GPT-5.4 Thinking, a new AI model designed for reasoning, coding, and enterprise workflows. The system is now available across ChatGPT, API, Codex, and the new ChatGPT for Excel integration.
OpenAI announced the release of GPT-5.4 on March 5, 2026. The new OpenAI GPT-5.4 AI model is now available in ChatGPT. In addition, developers can access GPT-5.4 through the OpenAI API and the Codex programming environment.
According to the company, GPT-5.4 combines advanced reasoning capabilities, software development tools, and agent-style workflows. As a result, the AI model is designed to support complex tasks across different digital environments.
Alongside the release of GPT-5.4, OpenAI introduced GPT-5.4 Thinking. This mode is intended to make AI reasoning more transparent. For example, GPT-5.4 Thinking can present an upfront plan before generating the final answer.
The company says this approach helps users understand how the AI model solves complex problems. Meanwhile, OpenAI also announced the beta launch of ChatGPT for Excel. The new tool integrates ChatGPT directly into Microsoft Excel workbooks and runs on the GPT-5.4 AI model.
OpenAI Releases GPT-5.4: What We Know About the New AI Model
According to OpenAI, GPT-5.4 is designed as a general-purpose AI model. It can handle tasks such as text generation, data analysis, software development, and workflow automation.
The GPT-5.4 model has been deployed across several OpenAI products at the same time. ChatGPT users can access GPT-5.4 directly through the assistant interface. Meanwhile, developers can integrate the GPT-5.4 AI model using the OpenAI API.
In addition, the Codex platform now runs on GPT-5.4 for programming-related tasks. As a result, the model can support software development workflows and coding environments.
A key addition in this release is GPT-5.4 Thinking. Unlike standard response modes, GPT-5.4 Thinking can generate a structured plan before producing the final output. Therefore, users can review how the AI intends to approach a task.
Key GPT-5.4 Features: Reasoning, Computer-Use Capabilities, and a 1M Token Context
OpenAI describes GPT-5.4 as a system that combines several emerging directions in artificial intelligence development.
- Reasoning and planning: GPT-5.4 Thinking can display the logical structure of a solution before generating the final response.
- Computer-use capabilities: In Codex, GPT-5.4 becomes the first general-purpose AI model with native computer-use functionality.
- Extended context window: Codex also introduces experimental support for a context window of up to one million tokens.
These capabilities allow the GPT-5.4 AI model to analyze larger codebases and documents. In addition, the model can interact with software tools and computing environments as part of automated workflows.
Where GPT-5.4 Will Be Used: ChatGPT, API, Codex, and ChatGPT for Excel
The GPT-5.4 model is being integrated across several OpenAI platforms and services.
Within ChatGPT, GPT-5.4 serves as a general-purpose AI model for writing, coding, and data analysis. Meanwhile, the GPT-5.4 Thinking mode is designed for tasks that require deeper reasoning.
Developers can also access GPT-5.4 through the OpenAI API. As a result, the AI model can be integrated into applications, automation tools, and enterprise workflows.
In Codex, GPT-5.4 functions as the core AI model for programming environments. Here, its computer-use capabilities and extended context window are being tested.
At the same time, OpenAI introduced ChatGPT for Excel. This tool embeds ChatGPT directly into Excel workbooks. As a result, users can interact with GPT-5.4 inside spreadsheets for financial analysis, data processing, and reporting tasks.

What GPT-5.4 Means for the AI Industry and Enterprise Software
The launch of GPT-5.4 reflects a broader trend in the artificial intelligence industry. Increasingly, companies are combining multiple AI capabilities into unified systems.
Until recently, reasoning-focused AI models, coding assistants, and productivity tools often existed as separate products. However, GPT-5.4 suggests a shift toward integrated AI platforms.
OpenAI appears to be building a single AI system capable of working across documents, software code, and structured data. As a result, enterprise users may rely on one AI layer for multiple workflows.
At the same time, the release of GPT-5.4 may increase competitive pressure on other AI companies. Many competing platforms still separate reasoning models, coding tools, and productivity integrations.
If successful, GPT-5.4 could become a foundation for future AI systems that combine data analysis, software development, and enterprise automation.



