OpenAI has announced a new update to its artificial intelligence lineup with the release of GPT-5.4. This version brings several upgrades, including stronger reasoning abilities, better coding support, and new automation tools.
One of the most interesting additions is a feature that allows the AI to interact with a computer environment. Instead of only giving instructions, the system can now perform certain tasks by operating software and digital tools through automated steps.
The new models are available through ChatGPT as well as OpenAI’s developer platform, making them useful for both regular users and software developers.
Also read: Google Introduces Gemini 3.1 Flash-Lite: A Faster and More Cost-Efficient AI Model
GPT-5.4 and GPT-5.4 Pro Versions
OpenAI has launched two versions of the new model:
- GPT-5.4
- GPT-5.4 Pro
The standard version appears inside ChatGPT as GPT-5.4 Thinking. It replaces the earlier GPT-5.2 Thinking model for some subscribers.
Currently, access to GPT-5.4 includes:
- ChatGPT Plus users
- Team plan subscribers
- Pro plan subscribers
The more advanced GPT-5.4 Pro model is available mainly for Pro and Enterprise users, which suggests it is designed for heavier professional workloads and complex tasks.
Better Reasoning and Coding Abilities
GPT-5.4 improves several core abilities compared with earlier versions. According to OpenAI, the model is better at understanding complex problems and generating useful solutions.
It also builds on earlier tools introduced in previous updates, including stronger programming capabilities.
With these improvements, GPT-5.4 can help users with tasks such as:
- Writing, reviewing, and debugging code
- Working with spreadsheets and documents
- Creating reports, presentations, and structured content
- Managing tasks that involve multiple steps or tools
These updates are meant to make the model more helpful for daily productivity as well as professional development work.
New Feature: AI That Can Use a Computer
A major highlight of GPT-5.4 is its computer-use capability. This feature allows the AI to operate inside a controlled virtual computer environment.
Using automated actions, the system can:
- Open and interact with applications
- Browse websites
- Perform mouse and keyboard actions
- Run commands on a system
- Complete multi-step workflows automatically
Instead of only explaining how to complete a task, the AI can generate the steps needed and carry them out within the environment.
This approach could make AI assistants more useful for tasks like organizing data, performing online research, or automating repetitive digital work.
Performance Improvements in Testing
OpenAI has shared testing results to show how the new model performs in different scenarios.
On the OSWorld-Verified benchmark, which measures how well an AI can navigate and use a desktop system, GPT-5.4 achieved about 75% success in completing tasks.
The model also performed well in visual reasoning tests. In the MMMU-Pro benchmark, designed to evaluate how well AI understands images and visual information, GPT-5.4 scored around 81%.
These results suggest the system is becoming better at interpreting visual elements and working inside graphical computer interfaces.
Additional Safety Measures
Because GPT-5.4 can interact with computer systems, OpenAI has added extra safety protections.
The company describes the model as having strong technical capabilities, which means safeguards are important when using it in real environments.
Some of the safety measures include:
- Tools that monitor the actions performed by the AI
- Restricted access for secure environments
- Systems designed to block unsafe or restricted operations
- New evaluation methods to review how the model reaches conclusions
OpenAI has also introduced methods to check whether the model clearly shows its reasoning process while completing tasks.
AI Assistants Are Becoming More Action-Oriented
Features like computer control show how AI technology is evolving. In the past, most AI tools focused mainly on answering questions or generating text.
Newer systems are starting to take action inside digital environments, which means they can assist with more complex workflows.
This development could help people automate routine work, organize digital tools, and complete multi-step processes more efficiently.
Also read: OpenAI Signals a Possible GPT-5.4 Release After Launching GPT-5.3 Instant
Conclusion
The launch of GPT-5.4 and GPT-5.4 Pro represents another step forward in OpenAI’s AI development. The update improves reasoning, coding support, and introduces the ability for AI to interact with computer systems in a structured way.
While some features are currently limited to specific subscription plans, the new model highlights a broader shift toward AI systems that do more than conversation. They are gradually becoming tools that can help manage and complete real digital tasks.
I am a passionate Tech Writer with strong industry experience. I enjoy exploring the latest technological innovations and sharing clear, helpful insights with my audience.”