I spent time testing Claude 4, Gemini 2.5, and o3

OpenAI's o3 did best on my quirky homebrew problems, with Gemini 2.5 Pro a close second.

A lot of models have come out in recent months:

• In February, xAI released Grok 3 and OpenAI released GPT-4.5.

• In March, Google released Gemini 2.5 Pro.

• In April, OpenAI released o3 and GPT-4.1, Meta released Llama 4, and Google released Gemini 2.5 Flash.

• Last week, Anthropic released Claude Opus 4 and Claude Sonnet 4.

Last year I made it a practice to do a w…