Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
MENU
OpenAI has launched GPT-5.4, positioning the new model less as a general chatbot refresh than as a sharper tool for professional output across documents, spreadsheets, presentations and software-driven workflows. The company says GPT-5.4 is rolling out across ChatGPT, the API and Codex, with an emphasis on higher-quality knowledge work, lower token use and fewer factual mistakes than GPT-5.2.
In practical terms, the GPT-5.4 knowledge work pitch is aimed at users who need more than polished prose. OpenAI is claiming stronger performance on structured business tasks such as spreadsheet modelling, presentation building and document editing, while also adding native computer-use capabilities for developers building agents. According to the company’s launch note, GPT-5.4 scored 83.0% on its GDPval benchmark for knowledge-work tasks, up from 70.9% for GPT-5.2, and reached 87.3% on an internal spreadsheet-modelling benchmark versus 68.4% for the older model.
That matters because OpenAI is clearly trying to move the discussion away from raw chatbot novelty and towards workflow insertion. The same release highlights improved results in computer use, browsing and tool selection, which points to a broader enterprise play: not just answering questions, but completing multi-step office and software tasks with less hand-holding.
OpenAI is also tying the launch to more concrete product hooks. Enterprise customers are being pointed towards a new ChatGPT for Excel add-in, while Codex and API users get updated spreadsheet and presentation skills. The company says GPT-5.4 supports up to 1 million tokens of context experimentally in Codex, although that sits above a standard 272K context window and comes with higher usage implications.
There is, however, a cost increase. OpenAI’s published API pricing puts GPT-5.4 above GPT-5.2 on both input and output tokens, suggesting customers will have to decide whether the productivity gain offsets the higher model price. That trade-off is likely to matter most in finance, legal work, operations and other document-heavy domains where accuracy and structure matter more than conversational flair.
The bigger point is that OpenAI is pushing harder into software used for routine knowledge work, not just AI experimentation. As previously reported by eeNews Europe when Capgemini partnered with OpenAI on enterprise AI agents, the company has been building towards deeper business integration. GPT-5.4 makes that strategy more explicit: better office outputs, better agent behaviour, and better integration with the tools people already use. For engineers, analysts and technical managers, the interesting question is no longer whether frontier models can write, but whether they can reliably do the drudge work around planning, modelling and cross-application execution.
All material on this site Copyright © 2026 European Business Press SA. All rights reserved.
