OpenAI Group PBC has launched GPT-5.4, a large language model (LLM) that enhances task automation capabilities compared to its predecessor, GPT-5.2. Available in ChatGPT and through OpenAI’s API, GPT-5.4 processes requests with up to 1 million tokens and uses significantly fewer tokens, lowering inference costs. A notable feature is its ability to automatically identify the necessary tools for applications, alleviating the need for lengthy tool lists that can inflate token usage. This model excels in handling images, supporting uploads of over 10 million pixels without compression. With a remarkable score of 75% on the OSWorld-Verified benchmark, GPT-5.4 outperforms prior models in user interface operations, spreadsheet analysis, and presentation preparation. Priced at $2.5 per million input tokens, the enhanced GPT-5.4 Pro edition offers maximum performance for complex tasks, signifying a major advancement in AI technology.
Source link
Share
Read more