New Developments

OpenAI has released two new models, GPT 5.4 Thinking and GPT 5.4 Pro. The GPT 5.4 Thinking model is specifically designed for enterprise work, such as coding and overseeing AI agents. This release is positioned as OpenAI’s most factual model yet, with responses being 18% less likely to contain errors compared to its predecessor, GPT 5.2.

One of the key features of GPT 5.4 is its support for agentic activity, which allows it to operate using less computing power, thereby reducing costs. OpenAI stated that this efficiency means it can more effectively support complex tasks.

Furthermore, the Thinking version of GPT 5.4 presents an upfront plan of reasoning, allowing users to make adjustments based on the model’s thought process. It supports up to 1 million tokens of context, significantly enhancing its usability for detailed inquiries.

Performance and Integration

Performance benchmarks indicate that GPT 5.4 outperforms earlier models, with OpenAI claiming it improves deep web research for specific queries. The Pro version is tailored for professionals who require maximum performance in their applications.

Additionally, GPT 5.4 is integrated into the API and Codex, expanding its accessibility for developers and enterprises alike. OpenAI has also secured a $200 million deal with the defense department, indicating the model’s potential applications in critical sectors.

Reactions and Future Outlook