← Back to the main Geek Mode article
What was tested: a single, simple request for a working, clickable stopwatch, given to Gemini 3.8 Flash (Flash Extended) and GPT-5.6 Sol (Extra High) in early September 2026. Outcome: Gemini answered with code to save and run separately. ChatGPT delivered a working stopwatch directly in the conversation.
The exact prompt
Make me a stopwatch so I can click start and stop on in this window with working buttons.
Gemini’s response
Gemini gave a perfectly reasonable block of HTML, CSS and JavaScript, followed by instructions to save it as an HTML file and open it in a browser. The code itself was fine. It simply was not what was asked for — a stopwatch usable right there in the window.
ChatGPT’s response
ChatGPT initially said it couldn’t guarantee that arbitrary live controls could run directly inside the conversation, then provided a working stopwatch anyway — Start, Stop, Reset and Lap buttons that worked right there, plus the HTML file as a backup.
Takeaway
Gemini answered the programming question. ChatGPT solved the problem. Knowing how to make something is not the same as actually completing the job — a distinction that shows up repeatedly across these tests.
← Back to the main Geek Mode article · Next: image generation and UI editing →