Choose a Claude plan by the interruption it needs to remove. A short session limit, a weekly allowance and an API budget solve different problems. Here is a dated comparison and a worksheet for deciding what extra capacity would be worth to you. Prices checked on 1 November 2026.

## Current individual plans {#plans}

[Claude's pricing page](https://claude.com/pricing) lists these USD web prices before applicable tax. Annual Pro costs $200 paid up front; monthly Pro costs $20. The table uses monthly billing throughout.

| Plan | Monthly price | Five-hour allowance | Weekly allowance |
| --- | ---: | --- | --- |
| Free | $0 | Limited | Limited |
| Pro | $20 | Base | Base |
| Max | $100 | 5× Pro | 5× Pro |
| Ultra | $200 | No limit for verified accounts | 10× Pro |

These allowances are not fixed message counts. [Anthropic's Ultra documentation](https://support.claude.com/en/articles/17728952-what-is-the-ultra-plan) confirms the verification condition and the continuing weekly limit. Removing the five-hour limit does not remove all limits.

## Identify the actual bottleneck {#limits}

[Anthropic distinguishes usage limits from conversation length](https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work). Claude chat, Desktop and Claude Code share usage. The account's Settings → Usage shows the session and weekly reset times. A context limit concerns one conversation, rather than the amount of work allowed over a week.

Record which limit interrupts you and the task you were trying to finish. Buying capacity for a different bottleneck may leave the interruption unchanged. For a week, keep one row per interruption: time, model, task, displayed limit, and whether the work actually had to wait.

## Put a value on uninterrupted work {#worksheet}

Use the difference in price as a hurdle, then test whether the upgrade removes your actual interruption.

| Monthly move | Additional cost | Time worth $20/hour needed to cover it |
| --- | ---: | ---: |
| Pro → Max | $80 | 4 hours |
| Max → Ultra | $100 | 5 hours |

The $20/hour figure is an illustrative value for your time, not a measured saving or a claim about earnings. Replace it with your own number. Four hours of blocked tasks are not automatically four hours recovered: you may have useful work to do while waiting, or the new plan may encounter another limit.

Before upgrading, write down one testable expectation, such as completing a particular daily workflow without a session interruption. Review the outcome after comparable work. Count completed deliverables and remaining interruptions, not just the size of the allowance.

## Keep API capacity separate {#api}

The pricing page currently includes $100/month of API credits with Max and $200 with Ultra. Follow the linked credit terms before budgeting: a credit balance and subscription usage are different quantities. For token-cost arithmetic, use the [API cost worksheet](/analysis/ai-api-pricing/); it includes cache writes, reads and retries.

For repository work, the [Claude Code setup and billing guide](/claude-code/) explains where to start and what to inspect in your account. If you are choosing between coding tools, use the [Codex, Claude Code and Cursor comparison](/analysis/codex-vs-claude-code-vs-cursor/#cost) to compare subscriptions, separate usage charges and review time.

Use the [Claude timeline](/claude/) for dated changes and the [model catalogue](/models/anthropic/) for model-specific facts. A product announcement explains what changed on that date; this reference is the place to compare the currently checked terms.

## Update the decision when the offer changes {#updates}

Recheck the provider links before paying. Update prices, conditions and the review date together. Keep the interruption log after a plan change so the next decision rests on your workload rather than a larger advertised multiplier.