← Back to context

Comment by Schlagbohrer

2 hours ago

250k tokens is awfully small for a frontier context window, I haven't pushed the frontier models in a long time but i assumed they all had 1M token windows as advertised practically a year ago.

They do, but compaction helps performance. All models degrade in performance the deeper the context. You can disable auto-compaction, but most of the time you want to /compact, at least, and likely /clear between discrete units of work.