Google says Gemini 4 Argon is the best coder alive, so I put it up against Claude Opus 5.5

Key Points:

• Google announced its latest AI model, Gemini 4 Argon, on September 30, claiming strong capabilities in complex software engineering tasks, despite the company's previous models often underperforming in coding.
• Unlike earlier Gemini models that rushed through tasks and delivered incomplete or buggy projects, Gemini 4 Argon demonstrated a thoughtful, personalized approach by building a sophisticated note-taking app tailored to the user's background and needs.
• The app created by Argon included advanced features such as an infinite canvas, knowledge graph, executable code blocks, spaced-repetition flashcards, and a fact-checking ledger, none of which were explicitly requested but showed impressive initiative.
• While the app had some initial technical issues like loading errors and refresh loops, Argon successfully identified and fixed these problems promptly, indicating a significant improvement in problem-solving and development process compared to previous Gemini versions.
• In comparison, Claude Opus 5.5 produced a more conventional, polished note-taking app with extensive testing and fewer bugs, but lacked the ambitious, customized features that made Argon's output stand out, marking Gemini 4 Argon as a notable advancement in Google's AI coding capabilities.