OpenAI Introduces GPT-5.4: A Major Upgrade in AI Capabilities
OpenAI has officially launched GPT-5.4, a new version of its artificial intelligence model, now available in ChatGPT. The company describes this model as its "most capable and efficient frontier model for professional work", marking a significant advancement in AI technology designed to enhance productivity and task automation.
Key Features and Enhancements of GPT-5.4
GPT-5.4 integrates improvements in reasoning, coding, and workflow automation into a single system. Building on earlier models like GPT-5.3-Codex, it offers enhanced capabilities for handling complex tasks. In ChatGPT, the GPT-5.4 Thinking feature allows users to view an outline of the model's plan before generating a final answer, enabling real-time instruction adjustments without restarting conversations.
The model also boosts internet search functionality, handling longer queries and maintaining context across multiple steps while gathering information from diverse sources. Additionally, GPT-5.4 introduces built-in computer-use abilities, enabling AI agents to interact with software, websites, and tools for tasks such as file management and cross-application workflows.
With support for up to 1 million tokens of context, GPT-5.4 can process large documents and extended conversations efficiently. In professional benchmarks, it achieved an 83% win or tie rate against industry experts in the GDPval test across 44 occupations, and scored 57.7% on the SWE-Bench Pro coding benchmark, showing slight improvements over previous models.
Availability and Deployment Details
OpenAI has begun a gradual rollout of GPT-5.4 starting today. In ChatGPT, GPT-5.4 Thinking is accessible to Plus, Team, and Pro users, replacing the older GPT-5.2 Thinking model, which will remain available for paid users for three months in a "Legacy Models" section before removal.
Enterprise and education customers can enable early access through administrator settings, while GPT-5.4 Pro is offered for Pro and Enterprise plans. Developers can access the model via the OpenAI API using the names gpt-5.4 and gpt-5.4-pro for higher performance tasks. The model is also integrated into Codex, providing coding and software development capabilities, with an experimental feature for application interaction during development tasks.
This launch underscores OpenAI's commitment to advancing AI tools for professional environments, aiming to streamline workflows and boost efficiency in various sectors.
