Thursday, January 8, 2026
AI Coding Model Comparison
People are comparing different AI models, including Claude Opus 4.5, for coding tasks and specific applications like tax filing and kernel work, highlighting their strengths and weaknesses.
Claude Opus 4.5 (w/ Claude Code) is known as the best coding model today, but which model is the best at filing taxes? We, at Column Tax, tested the latest crop of frontier models and here's how they stacked up on TaxCalcBench: - GPT-5.2 Pro: 41.18% fully correct returns -
here is my take on ai coding tools for kernel work: gemini 3 pro - best right now. most up to date and systematic in it's approach. opus 4.5 (cursor) - quite good but gives up easily. opus 4.5 (claude code) - not as good gpt-5.1-codex - needs hand holding but decent.