It’s highly unlikely that the Claude model itself is broken; the issue is more likely with the tool-calling layer.

at this point, i expect them to be all fine tunes of the same model with different prompts depending on you social class, foreign status and whether youre interested in competing with Anthropic.

no one can explain why AI wont be the worst of all its internet ancestors. its both in its training data and a pure capital wonder.

anyone hooking their livlihood to the AI revolution is either ignorant or sociopathic.