Can someone help understand what is this ‘non-max-claude-4.5-opus-high-thinking’ that I’m seeing under the usage. How is this different from the ‘claude-4.5-opus-high-thinking‘ that’s included?
Hey there.
On-demand usage splits model IDs by max mode state (non-max-) prefix for non-max requests), while included usage doesn’t make this distinction.
(gpt-5.1-codex-max is just a different variant name, not Max mode 
)