A hands-on comparison of Claude Sonnet 4.6, ChatGPT 5.5, and Gemini 3.1 on a zero-shot debugging task involving three deliberately introduced bugs in a Pygame platformer. Claude identified all three bugs including swapped platform momentum axes, conditional gravity logic, and inverted wall collision. ChatGPT found two out of three, missing the wall collision bug. Gemini failed to find any bugs and instead rewrote the control scheme entirely. The test highlights meaningful differences in how these models approach precise code analysis versus improvisation.

4m read timeFrom xda-developers.com
Post cover image
Table of contents
I put the "intuitive" intelligence of the models to test with 3 bugsChatGPT 5.5 uncovered 2 out of 3 bugsGemini 3.1 got 0 out of 3Claude Sonnet 4.6 uncovered all 3 bugs
132 Impressions