IBL News | New York
OpenAI released its new foundation model, GTP-5.4, last week, presented as “the most capable and efficient frontier model for professional work, involving spreadsheets, documents, and presentations.”
It’s also OpenAI’s first model with native computer-use-capabilities, enabling agents to operate computers and carry out complex workflows across applications.
In addition to the standard version, the San Francisco lab introduced GTP-5.4 Thinking as a reasoning model, and GTP-5.4 Pro for high performance.
The model’s API has been released with the largest context window from OpenAI: 1 million tokens.
Also, the company has reworked how the API manages tool calling, introducing a new system called Tool Search, resulting in faster and cheaper requests.
OpenAI said GPT-5.4 can write code to operate computers and issue keyboard and mouse commands in response to screenshots.
GPT-5.4 also showed improvements while using web browsers and gathering information from multiple sources, too, as the company says the model “can more persistently search across multiple rounds to identify the most relevant sources, particularly for ‘needle-in-a-haystack’ questions, and synthesize them into a clear, well-reasoned answer.”