Replying to Avatar The Fishcake (nostr.build)

After spending a ton of time vibe coding, here is my short review of the AI models.

- Claude Sonet 4.0 💩💩💩

- Cloude Opus 4.1 💩💩

- Grok mini - good for small changes or no BS refactoring

- GPT 5 mini - good for small feature additions and short context

- GPT 5 - good for large changes and complex optimizations with complex math and algorithms

- The rest - 💩💩💩💩
Avatar
aljaz 3mo ago

Are you basing the sonnet/opus scores on the last 1-2months since they were lobotomized by anthropic or before?

Reply to this note

Please Login to reply.

Discussion

Avatar
The Fishcake (nostr.build) 3mo ago

Last two week, and many months of use before that. They are a super convincing liars, those models