← Back to the main Geek Mode article
What was tested: practical questions about a new Anker charger and battery packs, then a request to crop, straighten and combine product and serial-number label photos into one documentation image with purchase-date context, given to Gemini 3.8 Flash (Flash Extended) and GPT-5.6 Sol (Extra High). Outcome: both systems got the important facts right; ChatGPT produced the cleaner final archival image.
Practical questions handled well by both systems included whether the battery could be fed through two USB-C cables to charge faster, which port to use, whether all three ports on the newer charger were equivalent, and whether the faint warmth noticed after leaving the packs connected overnight was normal.
For the documentation task, both systems were given photographs of the product label and the serial label and asked to crop and straighten the important areas, combine them into one image, look up the first day of Prime Day 2025, and add a small record showing the purchase date and first-use date. Both produced the right information and preserved the important product identifiers correctly. The difference was presentation: Gemini produced a functional composite with labels arranged vertically, looking like a quick collage. ChatGPT produced something closer to a proper reference sheet — straight horizontal labels, clean whitespace, a large readable information block. Neither system failed this test; one simply produced the artifact worth keeping.
Part of the serial number was intentionally obscured for public publication. Both original AI outputs reproduced the identifying numbers correctly.
← Previous: app development estimates · Back to the main Geek Mode article · Next: 18-page PDF test →