Recently, I was tasked with automating the generation of templates for simple resumes, and I tried various models. Here is my ranking of the different models based on the generated templates: purely subjective, mainly judging by structural completeness (as defined by JSON schema), AI/new technology content (to see if it uses the latest data/technology stack), and the depth of text details. I generated one in Chinese and one in English. The ranking is as follows: claude-opus-4.6 No doubt, very strong. The only one with a very advanced tech stack, featuring Vercel AI SDK, Claude Code, Codex Cursor, etc. gemini-3.1-pro Strong kimi-k2.5 Unexpectedly, I think it's the best in the country qwen3-max Content is quite detailed, fairly standard qwen3.5-plus Not bad MiniMax-M2.5 Average, outdated, still on Next.js 13+ glm-5 Really not good Also, some of GPT's models were not tested.