Transcription
I think Claude the chatbot is really C tier. Let's do an AI tier list for May 2026.
First up, we got Codex. I'm putting this S tier, right up there with Claude Code. These are the two best agent coding agents for a reason. And if anything, Codex is pretty underrated because you can get comparable outputs at a way lower price point than you can with Claude Code.
And speaking of Claude, let's talk about Claude the chatbot. Lots of people would assume this would be S tier, but I think Claude the chatbot is really C tier. And the reason for that is you're not going to get enough usage out of it at the $20 a month price point for it to be your daily driver. If you're on the $100 to $200 a month plan, sure, it's S tier. The models are amazing. But for your average user, you don't get video generation, you don't get image generation, and you have usage problems. Why would you do that?
So, if we're talking about what should you pay 20 bucks a month for, I think it's going to be either ChatGPT or Gemini. ChatGPT has been rising for me lately, in large part because their new image model is so good. Their new image model is frankly better than Nano Banana Pro.
And speaking of Nano Banana Pro, I'd put that at A tier. The new GPT Images 2 is solid S tier. It does text amazingly well. It's just an amazing product.
Then we got Hermes. Lots of hype around Hermes, but I would put it C tier, and it's basically just a better OpenClaw. And I think OpenClaw is D tier. These are both productivity theater tools. You can do everything inside of them. You can do inside of Claude Code or Codex for cheaper. People just get really excited about throwing wrappers around these things and feeling like they're on the cutting edge of AI.
Oh, hey, I almost forgot about Grok, and that's for a reason. Unless you're a Twitter brain, it is solid C tier. The models aren't that great in terms of like actual productivity. It doesn't do images well, doesn't do video well. All it really does well is current like events. And if I really cared about that so much, I would just use Perplexity, which is just a better Grok.
Then we got Naden. Solid B tier these days. It's pretty niche, but it's definitely a step above Zapier and Make, which are solid D tier for me. Why would you use these?
And we actually have a lot of residents of D tier lately. Honestly, Lovable these days. D tier. No reason to use Lovable. No reason to use Bolt. No reason to use Copilot, specifically with Bolt and Lovable. Why would you use these tools anymore when Codex exists, when Claude Code exists? It's totally eaten up its market share. You can do more at a cheaper cost.
Then we got Anti-Gravity. Solid B tier. It's fine. And it's a nice IDE if you really love Google products.
We got Seed Dance. Seed Dance definitely S tier. It's the best video model on the market right now. V3 B tier. Uh, we kind of need an update. Apparently Omni is going to be coming out soon. Ling 3.0 A tier, just a step below Seed Dance.
Then we got Cursor. Cursor I think I would put A tier. A lot of people really like Cursor. It gives you access to all the models and Composer 2.0 is actually solid.
And last but not least, we have Claw Design, which in terms of an actual tool, I would put S tier, but because of usage limits, which are like D tier status, I just put it solidly at B tier.