Claude Max vs Ultra: What the Extra $100 Buys

Max and Ultra side by side

Everything else matches: both include Claude Code, Opus, priority access at busy times and early access to new features, and both can continue past a limit on usage credits at standard API rates. Read the table by the two limit rows, because they are the only usage difference.

Ultra is the former Max 20x, now without a five-hour cap

On October 30, 2026 Anthropic renamed Max 5x to Max and Max 20x to Ultra. Prices stayed at $100 and $200, and subscribers moved to the new names without doing anything. Anthropic says weekly limits did not change. The change for Ultra is the five-hour limit: verified Ultra accounts no longer have one.

Verification is automatic. Anthropic checks factors such as account and payment history; until it does, a newer Ultra account keeps a temporary five-hour limit, and verifying a phone number at checkout does not remove it. Anthropic said it is lifting the limit for most existing subscribers first and emails each account when its limit is removed. Settings, then Usage tells you where you stand: if there is no Current session bar, the account is verified.

Which limit stops you decides the plan

Check Settings, then Usage on Max for a few weeks and note which limit you reach.

• The five-hour limit, with weekly room left at the reset. This is the case Ultra was changed for: a long Claude Code session or a deadline afternoon hits the five-hour cap while the week's allowance is far from used. Ultra removes that cap. The cheaper fix, if it happens only a few times a month, is usage credits on Max for those peaks.
• The weekly limit, before it resets. Ultra doubles it. Usage credits can bridge an occasional short week, but if Max's weekly limit runs out most weeks, Ultra's doubled allowance costs less than buying the shortfall at API rates.
• Neither. Stay on Max. Ultra's extra room goes unused, and only its API credit could still justify it (see below).

Usage credits are priced per token, so the comparison with Ultra's extra $100 can be estimated. As an illustration, not an Anthropic figure: one Claude Code turn on Opus 5.5 that sends 40,000 input tokens and returns 2,000 output tokens costs $0.16 + $0.04 = $0.20 at Opus 5.5's $4 and $20 per million tokens, so $100 of credits covers about 500 such turns. If your monthly overflow on Max stays well under that, Max plus credits with a monthly cap is cheaper than Ultra. If it regularly passes it, Ultra costs less and its weekly limit is twice Max's.

API credit narrows the gap only above $100 of API spend

Both plans include monthly API credit for the Claude API, the Agent SDK and claude -p run with the linked Console organization's key. It does not pay for interactive Claude Code or extra plan usage, and unused credit does not roll over. Max's $100 already covers the first $100 of API use, so Ultra's larger credit only matters once your API spend through that organization passes $100 a month:

The table is ToolColumn arithmetic from the two plan prices and credits. Read the last column: at $200 or more of monthly API spend, Ultra's larger limits cost nothing extra. Moving up mid-cycle is charged pro rata, and claimed API credits are prorated straight away, with the full $200 from the next cycle.

Fable room doubles with the weekly limit

Both plans let Fable 5 and Fable 5.1 use up to half of the weekly limit, and Fable draws that limit down faster than other models. Because Ultra's weekly limit is twice Max's, half of it is twice as much Fable work. If Fable is the model you reach for on long tasks and Max's Fable share runs out midweek, that alone can justify Ultra. On Pro, Fable runs only on usage credits.

No discounts, and when another plan fits better

Anthropic offers no standard discount on Pro, Max or Ultra, and its support team cannot issue coupons; occasional promotions are announced through its official channels. Max and Ultra are sold monthly only. Moving from an annual Pro plan credits the unused Pro balance to your account, provided the billing address on the new plan matches the old one.

For two or more people, Claude Team is the other plan type to price. A Premium seat costs $100 a month billed annually ($125 monthly) for five times a Standard seat's usage, runs Fable within its weekly limit and adds $100 of API credit per seat, pooled up to $500 a month for the team, with shared administration. Work that runs unattended belongs on the API rather than on either personal plan, and Claude usage limits by plan shows where Pro sits below both.