Anthropic does not publish an exact token count for the Claude Max 20x plan, so there isn’t a fixed official number to quote. Community estimates put it at roughly 20x the Pro baseline, with one estimate around 880,000 tokens per 5-hour window and another around 220,000 tokens per 5-hour window depending on how the baseline is measured.

What’s actually known

  • The Max 20x plan is described as 20 times the Pro usage budget , not as a fixed token allowance.
  • Anthropic keeps the raw token ceiling opaque, so third-party numbers should be treated as estimates rather than guarantees.
  • In practice, heavier models like Fable 5 can burn through that budget quickly, with user reports saying it may consume about 2% of the quota per minute.

Practical takeaway

If you want a safe mental model, think of Max 20x as very large but not unlimited usage, especially for long agentic sessions or heavy Fable 5 work. For routine tasks, the effective amount can feel much bigger than for long code-generation or reasoning-heavy runs.

TL;DR: there’s no official exact token number, but public estimates place Claude Max 20x somewhere in the hundreds of thousands of tokens per rolling 5-hour window, with behavior varying by model and task.[6][9][2]