> It's no coincidence the flat-fee service is visibly crippled, while per-request API users are not reporting any difference. (edited) If the model (gpt4) is u

> It's no coincidence the flat-fee service is visibly crippled, while per-request API users are not reporting any difference. (edited) If the model (gpt4) is unchanged, but they tweaked their system prompt for it on the chat interface, this is what you would see.