I went down a rabbit hole this morning reading the 2025 Juejin AI tool roundups back to back, and the thing that finally crystallized for me is that the gaming tier list has quietly colonized the AI tool review, and almost nobody is naming it as the phenomenon it is. The first search result I opened had Cursor at S, Claude Code at A, Replit and Chef at B, with explicit "闭眼选" and "强烈考虑" and "先观望" labels next to each row. I have seen the same S-A-B-C-D scaffolding in at least four different Juejin roundups in the last month, sometimes with an extra F tier, sometimes with plus and minus sub-grades, sometimes collapsed into "S档", "A档", "B档", "D档" with no C at all. It is the same visual grammar you see in every Genshin Impact character tier list, every League of Legends champion ranking, every Smash Bros matchup chart, just with Cursor and Claude Code and v0 standing in for the characters. To be fair I think the format works for the format's intended audience, and I want to put down why I am a little skeptical of it as a buying signal before I forget.
The piece that pushed me over the edge was the "2025 年 AI 编程工具评测" post, which laid out a clean S-tier for Cursor, an A for Claude Code, a B for Replit and Chef, and a D for the long tail, with bullet points for "技术开发者" or "非技术用户" next to every row. The reasoning is not bad — Cursor gets S for the community and the Claude Sonnet support, Claude Code gets A for the underlying model quality, Replit and Chef get B for being good-but-narrow. The shape of the analysis is reasonable. What I find interesting is that the tier labels themselves do not really mean anything measurable. There is no benchmark behind an S, there is no score range that maps to A, there is no reason a tool jumps from B to A other than the author deciding it has earned it. I have not stress-tested Chef or Replit the way I have with Cursor and Claude Code, so I would not oversell or undersell the comparison, but the tier-list scaffolding is a vibes format dressed up as a ranking format, and the vibes are doing all the work.
The contrast with the more grounded roundups is doing a lot of heavy lifting in my head right now. The 2025 AI tool pricing guide post from the same set of search results laid out ChatGPT Plus at twenty dollars a month, Claude Pro at twenty, Google AI Pro at 19.99, Grok Premium Plus at forty, Perplexity Pro at twenty, Midjourney at thirty, and just said what each one costs and what you get. The persona-specific list post said "if you are a front-end engineer, here are four tools" and then named them. The case study post said "I shipped production code with both of these, here is where each one broke." None of those posts needed an S or an A or a D, and I had not really noticed that the S-tier format was the odd one out until I read five of these in one sitting. Honestly I think the gaming tier list is a 2024 artifact the AI tool ecosystem has not quite outgrown, the same way the eight-tool scoring matrix was a 2024 artifact the roundup ecosystem is currently outgrowing.
The meta-pattern I want to put down before I forget it is that the S-A-B tier list is borrowing the visual authority of competitive gaming rankings to compensate for the fact that AI tools do not actually have a clean ranked order. In League of Legends there is a real patch-by-patch win rate you can point to. In Genshin there is a Spiral Abyss clear rate. For Cursor versus Claude Code versus v0 there is no equivalent number, and the tier list is the workaround. I am a little skeptical of any S-tier label that does not show the underlying benchmark, but I also get why the format exists. A reader who has never used any of these tools wants a one-glance answer, and the tier list gives them one. The second-time buyer wants the case study. The third-time buyer wants the pricing math. The S-tier post is the first-time-buyer's format, and the Juejin roundups are right to keep writing them, but I am going to keep skipping them in my own reading.
I will reassess in three months. The last time I said that I was mostly bouncing between Cursor and Claude Code, which is still where I land for coding, and reading the S-tier roundups is still useful for catching tools I had not heard of. What has changed is that I now read the S-A-B-D label as a flag to check the underlying reasoning rather than as a buying recommendation, and I think that filter is going to age well. Give it six months and the gaming tier list format might evolve into something with a real benchmark behind the letters, but for now the letters are doing all the work and I am done pretending an S is a measurement.
