OpenAI, in Desperate Need of a Win, Launches GPT-5.4
The new model can autonomously operate apps and issue keyboard commands, marking a major agentic AI leap.
OpenAI has released GPT-5.4, its latest flagship model, in a strategic move to regain public trust and user growth following significant backlash over its Department of Defense contract. The company frames GPT-5.4 as its "most capable and efficient frontier model for professional work," unifying advancements in reasoning, coding, and the critical new area of agentic workflows. The launch comes after the platform reportedly lost 1.5 million users due to the controversial DoD partnership, a decision that also sparked internal dissent. This release is clearly aimed at course-correcting with substantive technical improvements rather than just public relations.
The model's headline feature is native computer-use capability, a first for a general-use OpenAI model. This allows GPT-5.4 to autonomously operate across different applications on a user's machine by writing and executing code, as well as issuing direct keyboard and mouse commands. To validate these agentic abilities, OpenAI claims top rankings on the Mercor APEX-Agents, OSWorld-Verified, and WebArena Verified benchmarks. For general ChatGPT use, the company states GPT-5.4's individual responses are 33% less likely to contain errors than GPT-5.2, with an 18% overall reduction in mistakes and fewer hallucinations. GPT-5.4 Thinking is available for Plus, Teams, and Pro users, while GPT-5.4 Pro is accessible via API and for Enterprise/Edu subscribers, representing a major bet that superior autonomous task performance can reverse recent user attrition.
- First general-use OpenAI model with native computer control, enabling autonomous app operation and keyboard/mouse commands.
- Claims 33% fewer errors in individual responses vs. GPT-5.2 and leads key agent benchmarks like Mercor APEX-Agents.
- Launch follows loss of ~1.5M users after DoD partnership controversy, making this a critical trust-rebuilding release.
Why It Matters
Moves AI from a chat tool to an autonomous digital worker, fundamentally changing how professionals can delegate computer-based tasks.