OpenAI Unveils GPT-5.4 with "Computer Use" Capabilities and New Pro Tier for Enterprise Workloads
OpenAI launches GPT-5.4, featuring AI agents that can use computers, browse the web, and control software. Explore the new GPT-5.4 Pro tier and safety features.
By: AXL Media
Published: Mar 6, 2026, 10:40 AM EST
Source: The information in this article was sourced from gHacks

Autonomous Agents Bridge the Gap Between Text and Task
The launch of GPT-5.4 represents a fundamental evolution in OpenAI’s model capabilities, moving beyond text generation into the realm of digital action. The most notable feature of this update is the "computer-use" functionality, which enables AI agents to operate within a virtual computer environment. Unlike previous iterations that were confined to a chat interface, GPT-5.4 can now browse the web, fill out digital forms, place orders, and edit complex spreadsheets or presentations. By simulating mouse and keyboard inputs, the model acts as a digital assistant capable of interacting with standard software tools much like a human operator.
Enhanced Reasoning and Coding Integration
GPT-5.4 incorporates the specialized coding strengths of the previous GPT-5.3-Codex while significantly boosting its logical reasoning performance. OpenAI reports that the update is specifically tuned for software development and multi-step workflows that require the use of external tools. A new "steerable behavior" feature has also been integrated, allowing developers and professional users to more precisely guide the AI’s methodology when approaching intricate problems. This focus on "agentic workflows" ensures that the model can maintain coherence over long, complex tasks without frequent human intervention.
Introduction of GPT-5.4 Pro for High-End Performance
To accommodate the growing demand for enterprise-grade AI, OpenAI has restructured its access tiers with the introduction of GPT-5.4 Pro. This high-performance version is designed for intensive professional workloads and is available exclusively to Pro and Enterprise subscribers. Meanwhile, the "GPT-5.4 Thinking" model—which replaces the older GPT-5.2 version—is now the standard for ChatGPT Plus, Team, and Pro users. This tiered approach allows OpenAI to provide specialized compute resources to businesses requiring the most demanding agentic automation and deep-reasoning capabilities.
Categories
Topics
Related Coverage
- Anthropic and OpenAI Prepare Next-Generation Systems as AI Agents Reshape the Cybersecurity Battlefield
- OpenAI Designates London as Global Research Epicenter Following Launch of Most Advanced Coding Model
- Global Tech Leaders Unveil Groundbreaking Multimodal AI and Dedicated Hardware
- ION Group CEO warns of AI ‘substitution fallacy’ as data center energy demand doubles by 2030