Hacker News Reader

Comment by nouveaux | Hacker News Reader

Coca Cola would like to have a word with you.

These models respond differently and have their own "personality". Even in coding, there are people who swear by one model over the other. I know engineers who just stick with Claude and could not care to try Codex. For them, if it's not broken, why fix it?

View on HN

> Even in coding, there are people who swear by one model over the other

I just swear at the models. =P But jokes aside, I liked Claude Code and found it a big productivity boost for a month or two. Then the honeymoon phase slowly ended and I realized how much of its code I was rewriting myself. I don't use assistants anymore except to summarize changes for commit messages or PRs (and then I rewrite those summaries).

Not sure how many developers are like me, but I am very open to Claude, very open to Gemini, open to open source models (including gpt-oss), but am very reluctant to use frontier OpenAI models. The Microsoft distrust runs extremely deep, the browser authentication dance demanded of users for ChatGPT was the most extreme of the major frontier models, and early OpenAI API service stability was absolutely terrible. Llama had my back back then.

This is is no way dismissing your concern but I think this reinforces my point about branding. Whether or not Microsoft is handling AI in a responsible way, we don't trust them due to their poor practices on Window.

Reply on news.ycombinator.com