I Kept Copilot, ChatGPT, and Gemini Open for a Month. The One I Reach For Depends on the Hour.
The three big chatbots are close enough that the comparison misses the point. What decides it is which one fits the task in front of me, not which benchmark topped last week.
Everyone wants a winner in the Copilot versus ChatGPT versus Gemini fight, and the leaderboards rotate weekly, which should tell you the winner is not the point. I kept all three open for a month and the honest result is that each owns a lane, and the lane depends on the hour and the task, not on a benchmark from March.
Copilot, the Bing and Edge one, wins when the answer is on the web right now. Its grounding in search means it cites live pages and handles what just happened better than a model working from training data. For quick lookups inside a browser session it is the path of least resistance, though for deep writing it leans on the model behind it the same as everyone else.
ChatGPT wins when the job is a working session: long context, file uploads, code, and a conversational memory that survives the thread. For anything I am building or revising across turns, it stays open longest. The integrations and the canvas-style editing make it less a chatbot and more a bench, which is why it is my default for real work.
Gemini wins on the multimodal and the long document. Dropping a hundred-page PDF or a video and asking questions across it is genuinely strong, and the integration with Google Workspace means the context is already there for many users. Where it lags is the same-place-every-time feel of a tool that has not yet earned the default slot for open-ended writing.
My take: stop asking which is best and start asking which is nearest the task. Copilot for live web, ChatGPT for the build, Gemini for the document and the media. The people unhappy with one of them usually picked it for the job the other two do better, and the fix is not a better model, it is a better habit.