Post

@codevsdev @OpenAI @claudeai For raw code generation, the model matters. For real programming, the loop matters more: repo context, evals, diff quality, and how often it hallucinates less after the first fix.
English

@codevsdev @OpenAI @claudeai Claude is currently better for programming. It handles larger codebases, produces cleaner output, and makes fewer basic mistakes compared to ChatGPT.
English

@codevsdev @OpenAI @claudeai Claude really good at coding,and it doesn't patch
On the other hand GPT is good for almost every other thing else. @OpenAI should work on the persistent memory. It just forgets easily
English























