Today’s AI models, ranked S to F.
Daniel Miessler’s tier list for heavy day-to-day coding work. It changes when his mind does.
S tier
Claude Fable 5.2Anthropic
Gemini 4 NeonGoogle
Claude Opus 5.5Anthropic
A tier
GPT-6.2 AstraOpenAI
GPT-6.2 SolOpenAI
B tier
Claude Sonnet 5.5Anthropic
C tier
Claude Haiku 5.5Anthropic
D tier
F tier
What this list means
Usable or unusable for heavy day-to-day coding work.
SABCDFUsableUnusable
I’m building full paid applications with auth and paying users all day, shipping a massive amount of code. Heavy building stuff.
So the tiers here are pretty simple, they’re just my opinion on which models are legit usable for heavy day-to-day coding work and which ones are basically unusable.
A model down low here might still be great at plenty of other things.