GPT-5.4: Revolutionizing AI with Native Computer-Use Capabilities

OpenAI has announced the release of GPT-5.4, a major breakthrough in AI as the first general-purpose model with native computer-use capabilities, enabling AI agents to control computers and perform complex workflows across applications.

What's New in GPT-5.4

GPT-5.4 brings together OpenAI's latest advances in reasoning, coding, and agentic workflows into a single frontier model. It incorporates the industry-leading coding capabilities of GPT-5.3-Codex while improving how the model works across tools, software environments, and professional tasks involving spreadsheets, presentations, and documents.

Native Computer-Use Capabilities

The most exciting feature of GPT-5.4 is that it's the first general-purpose model with native, state-of-the-art computer-use capabilities, enabling agents to operate computers and carry out complex workflows across applications.

The model supports up to 1M tokens of context, allowing agents to plan, execute, and verify tasks across long horizons. GPT-5.4 also improves how models work across large ecosystems of tools and connectors with tool search, helping agents find and use the right tools more efficiently.

Key Point: GPT-5.4 is the most token efficient reasoning model yet, using significantly fewer tokens to solve problems compared to GPT-5.2.

Benchmark Results

GPT-5.4 sets new state-of-the-art across multiple benchmarks. On GDPval, which tests agents' abilities across 44 occupations, GPT-5.4 achieves 83.0% match or exceed rate compared to 70.9% for GPT-5.2.

Computer Use Performance

On OSWorld-Verified, GPT-5.4 achieves a 75.0% success rate, far exceeding GPT-5.2's 47.3%, and surpassing human performance at 72.4%.

On WebArena-Verified, GPT-5.4 achieves 67.3% success rate compared to 65.4% for GPT-5.2.

Visual Understanding Improvements

GPT-5.4's improved computer use is built on better visual perception capabilities. On MMMU-Pro, GPT-5.4 achieves 81.2% success rate, improving over GPT-5.2's 79.5%.

For Developers

For developers looking to use these capabilities via the API, they can be accessed using the updated computer tool. Users can also try these capabilities in ChatGPT using GPT-5.4 Thinking or Pro.

Conclusion

GPT-5.4 represents a significant step forward for OpenAI in developing AI that can perform complex tasks effectively. With native computer-use capabilities, AI agents can now truly work — not just answer questions or write code, but control and interact with software effectively.