Current AI usage ratings: GPT: 9/10 – The 'think' mode is already super useful: fast, accurate, and fresh. Claude: 8.5/10 – Probably lacks computing power, sometimes lazy and makes mistakes. Manus: 7.5/10 – Optimization is decent, but too expensive. Kimi: 7/10 – Multi-agent clusters are great, but lots of errors. Doubao: 6.5/10 – Weak capabilities, but good at handling emotions for the elderly and kids… GROK: 6.5/10 – Strong in X search, but uses a lot of computing power. Guessing @elonmusk will push it further later. Qianwen: 6/10 DeepSeek: 5/10 Hunyuan: 2/10 Wenxin Yiyan: 1/10 If I could only pick one, I’d recommend GPT. If Claude didn’t have the risk of being blocked, it could also be a main choice. These two are neck and neck—whichever gets updated next will likely surpass the other.