Claude 4.0 sonnet has politely answered my begging for the torturing back-and-forth hardship with long-thinking-but-mend-nothing of the “high-reason” GPT 5
1 Like
As far as I know, that’s a common misunderstanding AUTO doesn’t use any Anthropic or OpenAI model.
does
Can you prove it?
Because the name it shows doesn’t reveal anything!
It doesn’t match its behavior, not its speed, and in many aspects it’s better while in many it’s worse and I’ve built a lot of things with it.
I’ve definitely seen Auto’s behavior similar to GPT-4.1 and some of Claude’s. But the full list of options is secret.
I almost always have one of Claude’s corresponding to me. This is especially evident in Cashe Reads.
Because it’s trained on their output, so it looks similar — but think about it: why would they pay Anthropic money and give it to you for free?
Anthropic’s API is the most expensive, so there’s no reason they’d use their model for AUTO when there are so many decent models out there, like KIVI, Qwen, and others.
The “Auto” model isn’t free though. I am still being charged for it.
No, you’re mistaken! If you look at the bottom left, it actually shows you what you’re being billed for. AUTO is completely free — people just get upset about it for no real reason.
Prove it then. This screenshot is from the dashboard, I barely used the Auto model, therefore, the $0.11 only.
You’re looking in the wrong place what you’re seeing there is all your usage and how much it would have cost if they were charging you.
Look here here you can see what’s actually being used.
