OpenAI has been locked in a growing feud with Anthropic, which has escalated recently. Amidst this competitive landscape, OpenAI has made a significant advancement in its AI offerings.
New Developments
OpenAI released two new models, GPT 5.4 Thinking and GPT 5.4 Pro. The GPT 5.4 Thinking model is specifically built for enterprise work, focusing on tasks such as coding and overseeing AI agents.
This latest version is designed to be more accurate and capable of handling complex tasks. According to OpenAI, responses from GPT 5.4 are 18% less likely to contain errors compared to its predecessor, GPT 5.2. Additionally, individual claims made by GPT 5.4 are reported to be 33% less likely to be false.
Performance Enhancements
GPT 5.4 supports agentic activity and operates with less computing power, which translates to lower operational costs. OpenAI stated that this efficiency allows the model to provide more relevant answers while reducing the need for multiple interactions.
Furthermore, the Thinking version of GPT 5.4 showcases an upfront plan of reasoning, allowing users to make adjustments as needed. It supports up to 1 million tokens of context, enhancing its ability to manage extensive information.
Integration and Benchmarks
The new model is integrated into the API and Codex, with performance benchmarks indicating that GPT 5.4 outperforms earlier models. OpenAI positions GPT 5.4 as a step forward in making AI more reliable for professional work.
With a GDPval score of 83 for GPT 5.4, it significantly surpasses the scores of 70.9 for GPT 5.3-Codex and GPT 5.2. Additionally, the OSWorld-Verified score for GPT 5.4 stands at 75, compared to 47.3 for GPT 5.2.
As OpenAI continues to innovate in the AI space, observers are keen to see how these advancements will impact the competitive dynamics within the industry. OpenAI’s claims suggest that GPT 5.4 will enhance the reliability and efficiency of AI applications in various professional settings.
Details remain unconfirmed regarding the broader implications of these developments on OpenAI’s rivalry with Anthropic.
