Скачать или смотреть Claude vs Gemini vs Codex

Today we are building a simple to-do app, but we are going to do it in a fun way. We are going to compare three popular AI models for coding: Gemini from Google, Codex from OpenAI, and Claude from Anthropic. I have already installed all the command line tools we need.

Our plan is to give each AI the same prompt and see who can build the app the fastest and with the best code. We will run them side by side, watch how each one does, and see how they handle a curveball or two. First, I set up each model’s CLI with the right commands.

After a bit of trouble getting Gemini to accept the auto mode (I learned its flag is YOLO, which is cool), all three models were ready to go. We gave them the same basic prompt: build a to-do app in Python. Watching them work, Codex finished fast and saved the data in a JSON file, adding some nice features like undo.

Gemini took a minimalist approach, using only the sys library and fewer lines of code. Claude offered an interactive app and seemed the most user-friendly, plus it even tested itself. After running all the apps, I think Claude wins for the best user experience.

Codex had the most features and solid code, while Gemini scored points for being the simplest and using the fewest lines. In the end, each AI did well in its own way: Claude felt the most human and easy to use, Codex was feature-rich, and Gemini was simple and efficient.