linzhangrun

ChatGPT 3.5 only came out about three and a half years ago. Claude Code was released in February 2025. In one academic year, LLMs went from barely handling high school math to disproving unit-distance conjecture.

Every two months we seems to have a new SOTA now. What do you expect AI coding to look like four years from now?

Will it be like Go after AlphaGo, where even Ke Jie could only say "they no longer knew what they were fighting against"? Or will it just be like today, just stronger, with humans still driving the work and AI as a powerful assistant?

In the last one or two months, starting from DeepSeek V4 Pro, there are quite many low-price Chinese models coming out. Their performance looks more or less similar to me: Mimo V2.5 Pro, MiniMax M3, and the just released GLM 5.2, etc.

Which model are you using now? Why? What are the good and bad parts?

My main coding agents are CodeX-CLI and OpenCode (Harness seems to have some problems). I also use CodeWhale, Antigravity-CLI and OpenClaude as supplements (because of network issues, I don't really dare to use Claude Code). In some special cases, bashagt.

What coding agents are everyone using, or do you have any recommendations? New tools are coming out like waves now.