New ChatGPT 5.4 Model Is ‘Built for Agents.’ Will It Lure Back Claude Converts?
Already tired of ChatGPT 5.3 Instant, released Tuesday? OpenAI has a new family of models ready for you. The company released two new models on Thursday, named GPT 5.4 Thinking and GPT 5.4 Pro. GPT 5.4 Thinking is designed for enterprise work, such as coding and supervising AI agents.
ChatGPT 5.4 is a “reflective” model, meaning it takes a little longer to prepare your responses, but those responses should be more accurate and handle more complex tasks. It is intended for use by AI agents, which are robots capable of operating independently. OpenAI said that version 5.4 can support agentic activity more efficiently, meaning it uses less computing power and therefore costs less.
OpenAI calls GPT 5.4 “the most evidence-based model yet,” inadvertently highlighting the very real problem of when AI models hallucinate or make things up. OpenAI said this should be less of an issue with version 5.4, as its benchmark reports show that version 5.4 answers are 18% less likely to contain errors and individual assertions are 33% less likely to be false, compared to GPT 5.2. However, always fact-check what an AI tool tells you.
GPT 5.4 Thinking and Pro are now available to paid ChatGPT users and in the API (a developer tool). GPT 5.4 Thinking is also present in Codex, OpenAI’s coding application.
Learn more: AI Slop is destroying the Internet. These are the people who are fighting to save him
GPT 5.4 is a boost for OpenAI. Having an agent-centric model designed for power users willing to pay a monthly subscription like 5.4 certainly makes OpenAI seem like it’s taking a hit at Anthropic’s Claude.
Anthropic and OpenAI are locked in an escalating feud, recently kicked into high gear by Anthropic. Super Bowl commercials attack the advertising deployment of ChatGPT. But things went downhill from there, with recent reports showing Anthropic’s popularity is growing. Claude mobile apps have landed top spots in Apple and Google’s app stores, and AI users are filling online forums with advice on how to move your data from ChatGPT to Claude. One of the reasons why some users are making this change is the growing controversy between the two AI titans and the US government.
As a war breaks out in Iran and deals with the growing international crisis that follows, the War Department (formerly the Department of Defense) negotiates contracts with AI companies. The original deal was for Anthropic’s Claude, but it fell apart last week when Anthropic refused to allow the US government to use its AI to monitor citizens and support autonomous weapon systems.
OpenAI stepped into this void, with CEO Sam Altman clarifying this week that it would implement safeguards and not be made available to intelligence agencies like the NSA. The company previously announced that it had reached a $200 million deal with the Department of Defense in 2025. Many questions remain unanswered about how AI of any kind is used by government agencies and defense contractors.
