Hacker News new | past | comments | ask | show | jobs | submit
I don't know why but every time I have tried to use Anthropic models I've been disappointed. From the OpenAI stable, I use GPT 5.6 Luna and only switch to Sol 6.1 if it can't figure something out. When I don't need to use OpenAI models, I use Deepseek v4.1 Flash 0731. Every now and again I chuck a few tokens Anthropic's way but the models just chew through the $$ and never seem to be like "wow! so AI! very code!" to the extent that they appear to be worth the cost. MMMV I guess.
MMDNV - luna (6 now) is a workhorse!