Home / Technology / OpenAI Drops GPT 5.4: Smarter AI for Complex Tasks
OpenAI Drops GPT 5.4: Smarter AI for Complex Tasks
5 Mar
Summary
- New GPT 5.4 Thinking and Pro models enhance enterprise work and agentic activity.
- GPT 5.4 is 18% less likely to contain errors than previous versions.
- OpenAI secures $200 million deal with the defense department for 2025.
OpenAI has introduced its latest AI models, GPT 5.4 Thinking and GPT 5.4 Pro, aiming to enhance capabilities for complex tasks and enterprise applications. The GPT 5.4 Thinking model is particularly geared towards powering AI agents, offering more efficient operation for autonomous bots. This new generation is touted as OpenAI's most factual model to date, demonstrating an 18% reduction in response errors and a 33% decrease in false claims compared to GPT 5.2 benchmarks.
These advanced models are now accessible to subscribers of ChatGPT and developers through the API. The release signifies a strategic move by OpenAI, especially as competition with Anthropic's Claude intensifies. Reports indicate a growing user interest in Claude, with its mobile apps achieving high rankings.
The escalating rivalry is highlighted by a recent defense contract situation. Anthropic reportedly declined a US government deal involving citizen surveillance and autonomous weapons, leading OpenAI to step in. OpenAI CEO Sam Altman stated that safeguards would be implemented, and the models would not be accessible to intelligence agencies. OpenAI previously announced a $200 million contract with the defense department for 2025.
