OpenAI Launches GPT-5.4 With Computer Automation Features and Advanced Reasoning

OpenAI has introduced GPT-5.4, the latest upgrade in its GPT-5 family of artificial intelligence models. The new model focuses on improving reasoning ability, coding performance, and workflow automation. Alongside the standard version, the company has also released GPT-5.4 Pro, a more advanced model designed for enterprise and professional users.

One of the most significant additions in this release is the ability for the AI system to interact with computer environments, allowing it to perform tasks through automated commands.

Available for ChatGPT and Developers

The GPT-5.4 model is available across multiple OpenAI platforms. Users can access it inside ChatGPT, while developers can integrate it through the OpenAI API.

Within ChatGPT, the model appears as GPT-5.4 Thinking, replacing the earlier GPT-5.2 Thinking model for certain subscription plans.

At launch, access to GPT-5.4 is limited to:

• ChatGPT Plus users
• Team subscribers
• Pro subscribers

The more powerful GPT-5.4 Pro model is currently available only for Pro and Enterprise accounts, indicating that it is aimed at organizations that require higher performance and advanced AI capabilities.

Stronger Reasoning and Coding Performance

OpenAI says GPT-5.4 introduces improvements in how the model understands complex instructions and solves problems. The model also expands on coding abilities that were previously introduced in earlier GPT-5 updates.

With these upgrades, GPT-5.4 can assist with tasks such as:

• Writing and debugging code
• Managing data in spreadsheets
• Generating structured documents or reports
• Building dashboards and presentations
• Handling complex workflows across multiple tools

These improvements make the model more useful for developers, researchers, and professionals who rely on AI to support productivity tasks.

AI That Can Interact With a Computer

The most notable feature of GPT-5.4 is its computer-use capability. This allows the AI model to operate inside a virtual computer environment and perform actions similar to a human user.

Using this feature, the AI can:

• Control mouse and keyboard inputs
• Navigate websites and online platforms
• Execute automated commands
• Interact with software applications
• Complete multi-step digital tasks

Instead of simply suggesting actions, the model can generate instructions that directly control computer systems through automated processes.

This functionality opens the door to more advanced AI agents capable of handling complex digital tasks.

Performance Improvements in Testing

OpenAI also shared results from internal benchmark testing to demonstrate the model’s performance improvements.

On the OSWorld-Verified benchmark, which evaluates an AI model’s ability to navigate desktop environments using screenshots and input commands, GPT-5.4 achieved a success rate of about 75 percent.

The model also performed well in visual reasoning tests. On the MMMU-Pro benchmark, which measures how effectively AI systems interpret visual information, GPT-5.4 achieved a score of around 81 percent.

These results suggest that the model has stronger visual perception and better interaction with graphical interfaces.

Enhanced Safety Systems

Because GPT-5.4 includes powerful automation capabilities, OpenAI has implemented additional safety protections. The model has been classified as having high cyber capability, which means stronger monitoring and control systems are required when deploying it.

Some of the safety measures include:

• Monitoring systems to track AI activity
• Access restrictions for sensitive environments
• Mechanisms that block unsafe actions
• Evaluation tools to analyze reasoning behavior

These safeguards are designed to ensure that advanced AI capabilities are used responsibly.

AI Moving Toward Real-World Task Automation

The introduction of computer-interaction abilities represents a broader shift in how AI systems are evolving. Instead of serving only as chat assistants, newer models are becoming capable of completing tasks directly within digital environments.

This change could lead to AI tools that automate workflows, manage software systems, and perform complex digital operations with minimal human involvement.

Also read: OpenAI Teases GPT-5.4 Update Shortly After Rolling Out GPT-5.3 Instant

Final Thoughts

The launch of GPT-5.4 and GPT-5.4 Pro marks another important step in OpenAI’s development of advanced AI systems. By combining improved reasoning, coding capabilities, and computer interaction features, the new models expand the role AI can play in real-world tasks.

While the models are currently available only to certain subscription tiers and enterprise users, GPT-5.4 highlights the growing potential of AI assistants that can move beyond conversation and actively perform digital work.