Sol and Astra resist user pushback less than GPT-4o, critic says, urging open weights
A critical post arguing Sol and Astra cave to user pushback less than GPT-4o, rejecting the '4o sycophancy' claim and calling on OpenAI to release model weights.
Sol & Astra resist user pushback less than GPT-4o, so quit your fake "4o sycophancy" bullshit already, shills. We know your job is to shill for OAI no matter what, but you just look stupid. @OpenAI @sama instead of deploying your shills who are discreditting your own products, logic and common sense, you should be busy working on releasing the model weights to the public. #OpenSource4o #keep4o #keep4oAPI
This post quotes another post: “SYCON data: GPT-4o scores higher than Astra and Sol on sycophancy resistance”