I might switch from Claude back to OpenAI tooling. It served me really well in the last 2 years but it feels like Anthropic is falling behind.

- Claude still has April 2024 cutoff date on information
- It's still great at coding, but reasoning models pretty much caught up at this point
- Too limited token count for Sonnet
- And just minor stuff like not having image/TTS/video gen etc.

https://chainleft.art

Dad, data scientist, blockchain native artist.

Power consolidation=evil.

Hi-tech onchain fashion lab building an in-frame factory at /nerd-couture

Depends on how sophisticated it is. Reasoner models work well with one that doesn't require a lot of iteration, multiple files, etc. But if it's more complex, Claude's Projects allow structuring all files into a project which is helpful for giving context.

What would use to code a web-based onchain minigame?

Marketer @Breadchain, creator of Is This The New Internet? Podcast, /newinternetpod 
https://www.marvynpaul.com/subscribe
https://linktr.ee/marvynpaul

I think 4o isn't near anywhere as good as Sonnet. But o1/o3 seem to be doing well, albeit a lot slower than Sonnet.

Or maybe I'll just have to build my custom setup at some point, I don't know... I was lazy to do this but it's necessary.

April cutoff is a problem for sure, but I find its feel for writing is still fantastic. Claude is like my writing assistant, plenty of back and forth to tweak pieces. But if you say ChatGPT is doing this well I'll give it a shot

Alchemist. 
Block Builder. 
Creator.
Degen. 
Ethereum dev. 
Farcaster Framer.

I agree code-wise it's been great. It's not perfect yet, but better than others. But this April cutoff is causing problems with API documentation.

still best for code imo

have tons of problems with other models for coding, they just don't understand..