Claude Max 5x vs Max 20x — what the second hundred dollars actually buys
Two rungs of one plan. Max 5x is $100 a month, Max 20x is $200 — and across all 53 rows of Anthropic's own plan-comparison table, every other cell is identical. You are buying a multiplier, nothing else. Priced against Anthropic's own scale, 5x costs $20 per unit of Pro usage — exactly what Pro costs — while 20x halves that to $10. The step up is the only rung on the ladder that makes usage cheaper.
| Category | Winner | Margin |
|---|---|---|
| Monthly price · web subscription | AClaude Max 5x | $100 against $200 — Anthropic's help centre publishes both figures; its marketing page shows only "From $100" |
| Usage included · per 5-hour session, against Pro | BMax 20x | 20x Pro against 5x Pro — four times the headroom for twice the money |
| Cost per unit of usage · price ÷ the vendor's own multiplier | BMax 20x | $10 per Pro-unit against $20 — and $20 is exactly what a Pro subscription costs, so 5x buys headroom at no discount |
| Marginal cost of the step · what the extra $100 buys | BMax 20x | 15 extra Pro-units for $100 works out at $6.67 each, a third of Pro's own price per unit |
| Features · everything that is not usage | ·Identical | Every row of Anthropic's Free/Pro/Max 5x/Max 20x table matches between the two Max columns — apps, models, context, priority access, payment options |
| Model lineup · which models you can pick | ·Identical | Fable, Opus, Sonnet and Haiku on both, with Fable metered at 50% of weekly limits on either tier |
| Context window · chat and Claude Code | ·Identical | 1M tokens for Opus 5 and Sonnet 5 in chat, 1M in Claude Code for the newest models — stated for all paid plans, so the tier does not move it |
| Priority at peak · high-traffic access | ·Identical | Priority access at high traffic times is a Max feature, not a Max 20x feature — Pro does not have it, and both Max tiers do |
| Billing flexibility · monthly or annual | ·Neither | Max is monthly-only on both rungs; annual billing exists for Pro and Team, so neither tier can be discounted by paying up front |
| Overage · usage credits past the cap | ·Identical | Both can switch to usage credits at standard API rates, with a monthly spend cap and a $2,000 daily redemption limit |
| Risk of overpaying · unused headroom | AClaude Max 5x | 5x costs $1,200 a year against $2,400 — the cheaper mistake if your usage turns out to sit below the cap |
| Best overall | ·Depends | Cheaper per unit above, cheaper per month below. The line is whether you hit the 5x cap in a normal week, not an unusual one |
If you want the Max features without buying the ceiling.
- Half the outlay — $100 a month, $1,200 a year, against $2,400 for the tier above
- Every Max feature already included — priority access at high traffic times, higher output limits and early access to advanced features are Max benefits, not Max 20x benefits
- The same models — Fable, Opus, Sonnet and Haiku, with the same context windows the tier above gets
- A safety valve if you misjudge — usage credits at standard API rates turn an occasional heavy week into a one-off charge instead of a permanent $100 a month
- Cheaper to be wrong — unused headroom is the most common way to overpay for an AI subscription, and this is the smaller version of that mistake
- Prorated upgrade — moving to 20x mid-cycle is charged pro rata, so waiting costs you nothing
If you are already hitting the wall in a normal week.
- Four times the usage for twice the price — the only step on Anthropic's consumer ladder where the unit cost falls
- $10 per Pro-unit — against $20 on Max 5x and $20 on Pro itself; the marginal 15 units cost $6.67 each
- Cheaper than the overage it replaces — usage credits bill at standard API rates, and Opus runs $5 in and $25 out per million tokens
- Keeps the one-hour cache — prompt caching lasts an hour on a subscription but drops to five minutes once you are drawing on usage credits, so staying inside the plan is worth more than the sticker suggests
- Fewer interruptions — the session window still resets every five hours, but you are four times less likely to meet it
- Agent teams change the arithmetic — Anthropic's own docs put agent teams at roughly 7x the tokens of a standard session
| Aspect | Claude Max 5x | Max 20x |
|---|---|---|
| Plan priceindividual, web subscription | $100/month Anthropic's help centre publishes the figure; the marketing pricing page shows only "From $100" for the Max card. Mobile pricing may differ by app platform A wins | $200/month Same source, same caveat. The figure appears nowhere on claude.com/pricing itself — only in the Max plan support article |
| Usage includedper 5-hour session, relative to Pro | 5x Pro Described as five times more usage per session than Pro, shared across web, desktop, mobile, Claude Code and Cowork | 20x Pro Four times the 5x tier's session allowance, drawn from the same shared pool across every Claude surface B wins |
| Price per unit of usagemonthly price ÷ the vendor's own multiplier | $20 per Pro-unit $100 ÷ 5. That is exactly what one unit costs as a Pro subscription — and more than the $17 a Pro subscriber pays on annual billing | $10 per Pro-unit $200 ÷ 20. The only rung on Anthropic's consumer ladder where a unit of usage gets cheaper rather than merely more plentiful B wins |
| Marginal cost of the stepwhat the second $100 buys | Baseline The comparison point: five units of Pro usage for $100 | $6.67 per extra unit The step adds 15 Pro-units for $100. Judged on the margin rather than the average, it is the cheapest usage Anthropic sells to individuals B wins |
| Billing cyclecan you pay annually | Monthly only The Max plan is currently sold as a monthly subscription. Annual billing exists for Pro and Team, and one row of the plan-comparison table says Max offers it too — the help centre and the pricing FAQ both say it does not | Monthly only Identical. Neither Max rung can be discounted by committing to a year, which is why the annual figures below are simply twelve monthly payments |
| Annual outlaytwelve months, no discount available | $1,200/year The full year at list. Anthropic states it offers no standard discounted pricing on any paid plan, including Max 5x and 20x, and that support cannot issue one-off coupons A wins | $2,400/year The gap between the tiers is $1,200 a year — roughly what five Pro subscriptions cost, for fifteen Pro-units of extra headroom |
| Overagewhat happens past the cap | Usage credits Prepaid credits billed at standard API rates, with a monthly spend cap, optional auto-reload and a $2,000 daily redemption limit. Available on Pro and both Max tiers | Usage credits Identical mechanism and identical rates. The hidden cost is that prompt caching drops from a one-hour lifetime to five minutes once you are spending credits |
| Moving between the tiersupgrade and downgrade mechanics | Prorated upgrade Stepping up from 5x to 20x is charged pro rata for the rest of the billing cycle, so there is no penalty for starting low and moving later A wins | Same billing engine Coming from an annual Pro plan, the unused balance is credited to your account when it exceeds the Max price — but only if the billing address matches the old subscription |
| Published hour estimateshow much work that actually is | None published Anthropic no longer prints hour ranges per tier. Limits depend on conversation length and complexity, the model and the effort setting, so there is no fixed message or hour count | None published Same silence. The only figures you can act on are your own: run /usage in Claude Code, or open Settings and Usage on the web |
| Capability | Claude Max 5x | Max 20x |
|---|---|---|
| Monthly price | $100 | $200 |
| Session usage vs Pro | 5x | 20x |
| Cost per Pro-unit of usage | $20 | $10 |
| Session reset window | Every 5 hours | Every 5 hours |
| Weekly cap on top | ✓ Across all models, fixed reset day | ✓ Across all models, fixed reset day |
| Usage shared across surfaces | Web, desktop, mobile, Claude Code, Cowork | Web, desktop, mobile, Claude Code, Cowork |
| Claude Code | ✓ Included | ✓ Included |
| Claude Cowork | ✓ Included | ✓ Included |
| Claude Design and Claude Science | ✓ Both included | ✓ Both included |
| Model lineup | Fable · Opus · Sonnet · Haiku | Fable · Opus · Sonnet · Haiku |
| Fable access | 50% of weekly limits | 50% of weekly limits |
| Context window, chat | 1M on Opus 5 and Sonnet 5; 500K on the 4.x flagships | 1M on Opus 5 and Sonnet 5; 500K on the 4.x flagships |
| Context window, Claude Code | 1M on the newest models, no usage credits required | 1M on the newest models, no usage credits required |
| Higher output limits | ✓ A Max benefit | ✓ A Max benefit |
| Priority access at high traffic times | ✓ Max-only — Pro does not get this | ✓ Max-only — Pro does not get this |
| Early access to advanced features | ✓ Listed on the Max card | ✓ Listed on the Max card |
| Usage credits past the cap | ✓ Standard API rates | ✓ Standard API rates |
| Prompt cache lifetime inside the plan | 1 hour | 1 hour |
| Prompt cache lifetime on usage credits | 5 minutes | 5 minutes |
| Annual billing | ✗ Monthly only | ✗ Monthly only |
| Standing discounts | ✗ None on any paid plan | ✗ None on any paid plan |
| Payment methods | Credit card only — no ACH, no invoicing | Credit card only — no ACH, no invoicing |
| SSO, SCIM, audit logs | ✗ Team and Enterprise features | ✗ Team and Enterprise features |
| Model training on your content | Opt-out | Opt-out |
| Where you see what you have left | Settings › Usage on the web, /usage in Claude Code | Settings › Usage on the web, /usage in Claude Code |
| Published hour estimates per tier | ✗ Withdrawn | ✗ Withdrawn |
Claude Max 5x
The Max feature set at the lower of two prices — and at Pro's unit economics.
Strengths
- Everything Max has — priority access at high traffic, higher output limits and early access all arrive at this rung, not the one above
- $1,200 a year — half the commitment, and the cheaper direction to be wrong in
- Five times Pro's session allowance, drawn from one pool across chat, desktop, mobile, Claude Code and Cowork
- The same models and the same context — Fable, Opus, Sonnet and Haiku, with 1M-token windows on the newest models in both chat and Claude Code
- An escape hatch that scales down — usage credits at API rates cover an unusual week without turning it into a permanent $200 a month
- Free to change your mind — the upgrade to 20x is prorated for the remainder of the cycle
Weaknesses
- No discount for the volume — $20 per unit of Pro usage is exactly Pro's own price, and worse than the $17 an annual Pro subscriber pays
- The weekly cap still applies — the five-hour session window is not the only ceiling, and the weekly one resets on a fixed day assigned to your account
- Agent-heavy work eats it fast — Anthropic's docs put agent teams at roughly 7x the tokens of an ordinary session
- Credits cost more than they look — once you are spending them the prompt cache drops from an hour to five minutes, so repeated context gets re-billed
- Monthly only — there is no annual price to soften the outlay
Best for
- Daily users whose heavy days are occasional rather than routine
- Anyone stepping up from Pro who wants Max's priority access without the top price
- People who would rather pay a small overage bill than a large subscription
Max 20x
The only rung where a unit of Claude usage actually gets cheaper.
Strengths
- $10 per Pro-unit — half of what Max 5x and Pro both charge for the same unit of session usage
- $6.67 for each extra unit — judged on the margin, the fifteen units the step adds are the cheapest usage Anthropic sells an individual
- Four times the room, twice the price — the arithmetic favours the buyer for the first time on this ladder
- Overage avoided is money saved — credits bill at standard API rates, and on the current sheet Opus runs $5 in and $25 out per million tokens
- Keeps the long cache — staying inside the plan preserves the one-hour prompt cache that credit spending cuts to five minutes
- Right-sized for agents — parallel agents and long autonomous runs are the workloads that make 5x feel small
Weaknesses
- Twice the standing cost — $2,400 a year, with no annual discount and no coupon route
- Not one extra feature — every non-usage row of Anthropic's own plan table is identical to the tier below
- Still capped — the weekly limit applies across all models, and switching model does not restore a plan-wide limit
- Unverifiable in advance — with no published hour estimates, you cannot size the tier before buying it; you can only measure your own usage on 5x
- Easy to overshoot — if you rarely finish a five-hour window on 5x, the second $100 buys headroom you never touch
Best for
- Engineers running Claude Code or agent teams through most of a working day
- People who already pay a recurring overage bill on Max 5x
- Anyone whose work stalls the moment a session limit lands
You just hit a limit for the first time
You are on Pro, you met the session cap once during a long refactor, and the upgrade prompt appeared.
Reasoning: One wall is not a pattern. Max 5x costs the same $20 per unit of usage that Pro does, so stepping up buys room without buying value — and if the heavy day was a one-off, usage credits at API rates would have covered it for a few dollars. Turn credits on, set a monthly cap, and let a month of data decide.
Claude Code is your working day
You run agents from morning to evening, on a large repository, and the five-hour window lands most days before you are done.
Reasoning: This is what the 20x rung is for. Four times the session allowance for twice the price is the only place on the consumer ladder where the unit cost falls, and hitting the wall daily means you are already paying for the difference in overage — at API rates, with the prompt cache cut from an hour to five minutes while credits are in play.
You want Max for the priority, not the volume
Your usage is moderate but your timing is bad — you work during peak hours and you want the queue to move.
Reasoning: Priority access at high traffic times is a Max benefit that arrives at the 5x rung. It is not one of the things the 20x tier adds, because the 20x tier adds nothing but usage. Buy the cheaper Max and keep the other $100.
You run agent teams and parallel subagents
Your workflow spawns several Claude instances at once rather than one long conversation.
Reasoning: Anthropic's own documentation puts agent teams at roughly seven times the tokens of a standard session, because each teammate carries its own context window. That multiplier lands directly on your plan limits, and it is the single most reliable way to exhaust the 5x tier. If this is your normal week rather than an experiment, buy the headroom.
You are budgeting a year in advance
You need one number for a finance form, and you would like to pay up front for a discount.
Reasoning: There is no discount to find. Max is monthly-only on both rungs, Anthropic states it offers no standard discounted pricing on any paid plan, and support cannot issue coupons. The year is $1,200 or $2,400 flat. If a single annual invoice matters more than the tier, Pro at $200 up front or a Team plan are the only routes to one.
Frequently asked.
Common questions about this comparison, with sources where they matter.
Q · 01 What do I get on Max 20x that Max 5x does not have? +
Q · 02 Is 20x really four times 5x? +
Q · 03 How many hours of Claude Code does each tier give me? +
/usage in Claude Code — press d or w to switch between the last 24 hours and the last 7 days — or open Settings and Usage on the web. Note the Claude Code figures are approximate and computed from local session history, so a second machine's usage will not appear.