I’m a long-time Plus subscriber and technical power user who primarily interacts with ChatGPT on desktop (Windows) using Chrome, Edge, and Firefox, as well as the official Electron-based app.
The desktop experience is catastrophically poor.
Responses often take 30 to 60 seconds to begin, especially when tools or image generation are involved. Token output is clunky and arrives in bursts, not as a continuous stream. In many sessions, the interface freezes entirely or becomes unresponsive. This includes full lockups where the system displays “ChatGPT is not responding.”
None of these issues occur in the mobile app. It works flawlessly under the same conditions.
Meanwhile, Grok and Gemini—despite using similar web technologies—offer a much smoother and more usable desktop experience. This clearly isn’t a JavaScript problem; it’s a design issue in OpenAI’s client-side architecture.
If this continues, OpenAI will lose desktop power users to competitors. This isn’t just a UX concern. It’s a long-term strategic risk. Power users produce high-quality prompts, structured feedback, and exploratory queries that reinforce the learning loop. Lose them, and model refinement suffers.
Please escalate this beyond tier-one support. This is not a bug. It’s a systemic blind spot.