An AI enthusiast conducted a comparison test between GPT 5.2 and Claude Sonnet 4.5, discovering that GPT 5.2 performed poorly in programming tasks. When developing an Electron media player project, it encountered multiple errors and required Claude’s assistance to fix them. The author pointed out that GPT 5.2’s coding capabilities don’t match its marketing claims, while Claude proves more reliable in solving technical problems. Meanwhile, GPT 5.2’s unique advantage is its video calling feature in the app. The test results reveal the gap between AI models’ actual performance and their promotional claims, providing practical insights for developers.
