Claude Max 5x vs Max 20x: when the upgrade pays off

Already paying for Max and wondering if 20x is worth another $100 a month? Both tiers share every feature - Claude Code, Cowork, priority routing. The only thing 20x buys is roughly four times more volume before you hit a wall. Here is the exact usage pattern that tells you whether you need it.

Update, 2026-09-19: Max plans no longer run a separate weekly limit scoped to Opus. Opus and Sonnet are both current, fully available models, and both draw from the single weekly limit that covers every model. The second weekly meter on Max is scoped to Fable instead: Max includes Fable for up to 50% of the weekly limit, after which Fable keeps running on pay-as-you-go usage credits. Pro has one weekly limit across all models and bills Fable from usage credits from the first message. Checked against a live Claude Max 5x account that used Fable, Opus and Sonnet in the same week: Settings > Usage shows current session, this week, and Fable this week, with no Opus meter.

Update, 2026-09-18: Both Max tiers now include Anthropic's Fable models (Fable 5, and Fable 5.1 since 2026-09-01) for up to 50% of the weekly limit. That 50% is a ceiling inside the same weekly allowance, not extra capacity, and Fable drains it faster than Opus or Sonnet; past the ceiling Fable runs on pay-as-you-go usage credits. If Fable is your daily model, the 20x budget matters more than it did when Opus was the heaviest option.

The short answer

What Max 5x and Max 20x actually include

Per Anthropic's Max plan documentation, both tiers are identical in every way except the usage multiplier:

Neither tier unlocks a capability the other lacks. Max 20x is not "more features" - it is the same plan with a bigger fuel tank. That reframes the decision entirely: you are not asking "which plan is better," you are asking "how much volume do I actually burn in a week."

How much more volume 20x actually buys

Anthropic does not publish the absolute message or token counts for either tier, only the multiplier relative to Pro. Independent community testing - not official figures - gives a rough sense of scale for Claude Code usage specifically:

*Approximate, workload-dependent figures from independent testing (not Anthropic's own published numbers). Anthropic has changed the underlying 5-hour session cap before - it doubled the Claude Code limit for Pro, Max, and Team plans on 2026-05-06 - so treat the ratios as more durable than the absolute counts.

The practical read: Max 5x already covers a genuine full working session for most developers. Max 20x exists for the minority of users who run past that - long unattended agentic runs, multiple concurrent Claude Code instances, or a full workday of continuous use rather than a single focused session.

The one signal that decides it

Skip the abstract "which plan is better" question. There is a single concrete test:

Open Settings > Usage on claude.ai and watch your weekly all-models percentage for two weeks. If it drops to zero before Thursday - meaning you are locked out of Claude for the last day or two of every rolling week - Max 20x will pay for itself. If it never drops below 20-30% by the time your week resets, you are paying for volume you are not using, and the extra $100 a month is better spent elsewhere.

This mirrors the same "measure first, upgrade second" rule that applies to the Pro-to-Max-5x decision, covered in Claude Pro vs Max: upgrading on a hunch, without checking whether you actually exhaust your current tier, is the most common way to overspend on Claude.

A decision tree by weekly Claude Code hours

If you would rather reason from your own workload than watch a usage meter for two weeks, this is the pattern that has held up across the independent testing and community reports cited above:

• Under 10 hours a week, occasional Claude Code sessions. Max 5x has more headroom than you will use. Stay on 5x.
• 10-25 hours a week, one focused Claude Code session most weekdays. This is the Max 5x sweet spot - roughly 6-8 hours of continuous runway per session, per the table above, comfortably covers a normal working day.
• 25-40 hours a week, Claude Code running most of the workday, every weekday. Watch your weekly percentage. If you are hitting the wall by Thursday, move to Max 20x.
• 40+ hours a week, or running multiple concurrent Claude Code instances / long unattended agent sessions. Max 20x is very likely the right starting tier - waiting out weekly lockouts at this volume costs more in lost time than the $100 difference.

Solopreneurs and small teams billing client hours should run the same math in dollars: if a weekly lockout costs you even 90 minutes of billable work, and your time is worth more than roughly $65 an hour, the extra $100 a month for Max 20x is already cheaper than the lost hours it prevents.

Opus strategy: 5x forces discipline, 20x doesn't

Neither Max tier gives Opus a budget of its own: Opus draws from the same weekly limit as Sonnet and Haiku, so 20x simply makes that shared limit larger. (Max plans used to run a separate weekly Opus cap; that meter no longer appears.) The practical difference is in how much you have to think about it:

On Max 5x, most users need to be deliberate - reserve Opus for genuinely hard reasoning tasks (architecture decisions, tricky debugging, dense analysis) and default to Sonnet for routine work, or the weekly pool runs dry mid-week. On Max 20x, the Opus allowance is large enough that heavy users can reach for it more casually without tracking consumption closely.

If your main reason for considering 20x is "I want to use Opus more freely" rather than "I keep hitting the overall weekly wall," it is worth checking your actual Opus usage in Settings > Usage first - many users who feel Opus-constrained on 5x are closer to comfortable than they assume once they look at the real percentage.

When 20x is not worth it

Skip the upgrade to Max 20x if any of these describe you:

• You have never seen your weekly all-models percentage drop below 30% before the reset.
• Your Claude Code usage is one session a day, a few days a week.
• You are upgrading because Max 20x "sounds like more" rather than because you have hit a documented wall on Max 5x.
• You occasionally hit the 5-hour session cap but your weekly pool is nowhere close to empty - that is a session-pacing issue, not a volume problem, and more weekly budget will not fix it.

The downgrade path is free and immediate to request (it takes effect at your next billing cycle), which makes a one-month trial of Max 20x a low-risk way to test the real question - do you actually consume the extra volume - rather than guessing from the tier names alone.

How to make the call this week

1. If you are on Pro, do not jump straight to comparing 5x vs 20x - work out first whether you need Max at all. See Claude Pro vs Max.
2. If you are on Max 5x, check Settings > Usage today, then again in a week. Two consecutive weeks of hitting zero before Thursday is the signal to upgrade.
3. If you are unsure how close you are running, put a live usage meter in your menu bar so the percentage is ambient instead of a page you have to remember to open. That alone often resolves the decision within a week, because you notice the pattern instead of guessing at it.