I get the stuff with the rate limits. What I’m wondering about is claims that the context windows even in max mode are truncated on any of the models. Can you confirm that all max models use the maximum context windows? Why do people using diagnostics imply otherwise?
Also, there was talk of more visibility into context that gets sent to the model. We all understand normal requests have truncation, but for max mode, as professional devs, we cannot adjust our prompting with such uncertainty into what the model sees. If you give use full context, why not tell us what’s being sent? This whole thing is quite odd…I don’t want to publicly speculate more than that.
At least give us an indicator if a full file is sent over or being truncated.
This is the only thing keeping me from subbing currently.