Claude Code devs: use previos model, compact at 200k tokens to avoid usage limit

1 comments

• $jazz9k 12 hours ago

  I see token burn as the limiting factor with AI.