ChatGPT 5.4 Pro Adds Native Desktop Control for Real-Time Work
TL;DR
ChatGPT 5.4 Pro now features native desktop control, allowing the model to interact directly with running applications and live workflows. According to AI Grid, the model hits a 52% success rate on professional task benchmarks, covering complex scenarios in finance and healthcare. On the Frontier Math benchmark, 5.4 Pro solves advanced mathematical problems that have consistently tripped up earlier AI models.
Nauti's Take
OpenAI will surely market that 52% benchmark figure as a breakthrough – but it also means nearly every other professional task still fails. Desktop control is nonetheless the right direction: real automation has to happen where people actually work, not inside abstract API sandboxes.
The big open question is how OpenAI handles privacy and security when a model gets full access to a live machine. The Frontier Math callout in the same breath as desktop control feels oddly placed – classic benchmark showrooming to juice the news cycle.
Briefingshow
Native desktop control marks the step from chatbot to genuine AI agent: the model no longer just reads text but acts directly within the operating system. A 52% success rate on professional benchmarks may sound modest, but for multi-step, high-complexity tasks it signals real progress. For sectors like finance and healthcare, this means routine work – data entry, report generation, research – could be largely automated using existing desktop software, with no custom integration required.
Video
Sources