New Model Family Launch

OpenAI introduced two additions to its model lineup: GPT‑5.4 Thinking and GPT‑5.4 Pro. The Thinking variant is described as built for enterprise tasks such as coding and supervising AI agents, while the Pro version targets power users willing to pay a subscription fee. Both models are marketed as "thinking" models, meaning they take a bit longer to generate responses but aim for greater precision and the ability to manage more complex queries.

Performance and Factual Accuracy

OpenAI positions GPT‑5.4 as its "most factual model yet," highlighting benchmark results that show responses are 18% less likely to contain errors and individual claims are 33% less likely to be false when compared with the earlier GPT‑5.2. The company emphasizes that the model is better suited for AI agents—autonomous bots that can operate with minimal human oversight—by using less computing power, which should lower operational costs.

Availability and Competitive Landscape

The new models are immediately accessible to paying ChatGPT subscribers and are also offered through OpenAI’s API for developers. GPT‑5.4 Thinking is additionally available within Codex, OpenAI’s coding‑focused application. OpenAI sees the release as a strategic counter to Anthropic’s Claude, whose mobile apps have recently topped the charts on both Apple’s and Google’s app stores. Users have been discussing migrations from ChatGPT to Claude, a shift partly driven by ongoing rivalry between the two AI firms and broader governmental scrutiny.

Government Contract Shifts

In parallel with the product launch, the U.S. Department of Defense (formerly the Defense Department) has been renegotiating AI contracts. An initial agreement with Anthropic fell apart after Anthropic refused to permit the U.S. government to use its AI for citizen surveillance or autonomous weapon systems. OpenAI stepped into the void, with CEO Sam Altman stating that the company would implement safeguards and would not make its technology available to intelligence agencies such as the NSA. OpenAI previously announced a $200 million deal with the defense department slated for 2025, though many questions remain about how AI from any vendor will be employed by government and defense contractors.

Implications

The rollout of GPT‑5.4 underscores OpenAI’s focus on higher factual reliability and on serving enterprise customers that need robust AI agents. By positioning the new models against Anthropic’s Claude, OpenAI is intensifying competition in the rapidly growing generative‑AI market. At the same time, the shifting defense contracts highlight how government demand for AI capabilities is influencing corporate strategies and prompting discussions about ethical safeguards and agency access.

Used: News Factory APP - news discovery and automation - ChatGPT for Business