OpenAI unveils GPT-5.4, a model presented as more efficient for professional work and complex task automation. The company emphasizes in particular its agent capabilities: AI can use a computer directly, navigate through software, and orchestrate actions on behalf of the user.
The major players in AI have only the word 'agent' on their lips, asserting that these are capable of performing a series of actions on behalf of the user, such as booking a flight.
The OpenAI AI can now use a computer
GPT-5.4 takes the bet to do better. The new model launched by OpenAI is the first of its kind to be able to directly use a computer, clicking on interfaces, typing text and using a mouse, reading screenshots or navigating through websites or software. Agents labeled GPT-5.4 would therefore be able to perform complete tasks such as filling out forms, sending emails, managing calendars or manipulating applications.
GPT-5.4 is more efficient at chaining multiple tools (which it can dynamically retrieve) and parallelizing actions. In fact, it should be better at orchestrating workflows, processes, and actions.
That's not all. OpenAI also announces that its new model is the most capable of performing complex professional tasks: creating spreadsheets in spreadsheet programs, presentations, or even structured documents. The company states that GPT-5.4 matches or exceeds human professionals in 83% of cases according to the GDPval benchmark, which simulates real tasks in 44 professions (finance, sales, healthcare, etc.). That's 71% for GPT-5.2.
Ā© OpenAI The model is also intended to be more reliable: erroneous statements and hallucinations would be 33% less frequent, and complete responses would contain 18% fewer errors. Developers will benefit from the capabilities of GPT-5.4, which integrates and improves upon those of GPT-5.3-Codex, the code-specialized model. In an example provided by OpenAI, GPT-5.4 can create a complete amusement park management game in the browser and then use tools to test it automatically.
Research is announced as more efficient, allowing the model to traverse multiple sources and synthesize information more effectively. The Thinking (advanced reasoning) variant of GPT-5.4 is currently being deployed in ChatGPT (for paying subscribers); for quick responses, it's GPT-5.3 Instant that takes over.
šš» Follow real-time tech news: add 01net to your Google sources and subscribe to our WhatsApp channel.
Source : OpenAI
- Google Play Store
- 4K Video Downloader Plus
- VLC Media Player
- Adobe Reader
- uTorrent
- WinRAR
- Xender
- Google Chrome
- CCleaner
- Mozilla Firefox
- Avast Gratuit
- ChatGPT
- Microsoft 365
- Opera One
- Avast One










- Google Play Store
- 4K Video Downloader Plus
- VLC Media Player
- Adobe Reader
- uTorrent
- WinRAR
- Xender
- Google Chrome
- CCleaner
- Mozilla Firefox
- Avast Gratuit
- ChatGPT
- Microsoft 365
- Opera One
- Avast One



