Cursor models randomly output in Chinese and Russian

[Artemonim]

Just don’t use Auto

How To Optimize Your Usage: The Best AI Models to Use, version 2.2

Previous version of the guide: Gemini 2.5 Pro + Claude 4 Sonnet Thinking > Grok 4 >? GPT-5 >> Auto > GPT-5 Mini > Sonic Grok 4 broken - I like his approach, but he is always unable to continue chat. Gemini 2.5 Pro less broken, then Grok 4 - It can also interrupt execution in the middle of a task, like Grok, but less often. Claude Sonnet 4 Thinking very good when he has enough intelligence to do the task, but expensive. Very good as QA Engineer. I usually use it when other models have ■■■■■■…