Claude Code pricing in 2026: plans, API rates, and usage limits
Published
Summary
Claude Code is included in every paid Claude plan: Pro at $20 a month, Max at $100, Ultra at $200, Team seats from $20, and Enterprise at $20 a seat plus usage. It isn’t included in the Free plan. You can also pay per token with an API key. This guide covers every plan, the API rates, how usage limits work, why usage climbs, and how to run Claude Code in the cloud. Prices are from Anthropic’s pricing page and Claude Code documentation, checked on November 6, 2026.
At a glance
Claude Code plans and prices
Every paid Claude plan includes Claude Code. The plans differ in how much usage they include. Usage is shared with Claude chat, so a long chat session uses the same allowance as a coding session.
| Plan | Price | Claude Code | Usage | Limits |
|---|---|---|---|---|
| Free | $0 | Not included | Limited chat usage | Five-hour and weekly limits |
| Pro | $20/month, or $17/month billed annually ($200 a year) | Included | The baseline | Five-hour and weekly limits |
| Max | $100/month | Included | 5× Pro | Five-hour and weekly limits, plus a separate weekly limit for Fable |
| Ultra | $200/month | Included | 10× Pro each week | Weekly limit only; verified accounts have no five-hour limit |
| Team Standard | $25 per seat a month, or $20 billed annually | Included | More than Pro | Per-seat five-hour and weekly limits; 2 to 150 seats |
| Team Premium | $125 per seat a month, or $100 billed annually | Included | 5× a Standard seat | Same as Standard |
| Enterprise | $20 per seat a month billed annually, plus usage at API rates | Included | Billed by usage | 20-seat minimum |
Ultra replaced the $200 Max 20x plan
On October 30, 2026, Anthropic replaced Max 20x with Ultra at the same $200 price. Ultra has 10× Pro’s weekly usage, and verified accounts have no five-hour limit, so heavy users can spend the week’s allowance in fewer, longer sessions. Max is now a single $100 plan with 5× Pro usage.
Free options
Is Claude Code free?
The Claude Code software is free to install, but you need a paid plan or API credit to use it. The Free Claude plan doesn’t include Claude Code. The cheapest options are:
- Claude Pro, at $20 a month or $17 a month billed annually. This is the lowest flat price.
- An API key from the Claude Console, billed per token at the rates below. With light use this can cost less than Pro, but there is no cap unless you set a workspace spend limit.
- Amazon Bedrock, Google Cloud, or Microsoft Foundry, billed per token to your cloud account. This suits companies that already have cloud credits or commitments.
Pay per token
Claude Code API pricing
With an API key, Claude Code bills every token at Anthropic’s list prices. Cache reads are what Claude Code pays when it re-sends the conversation it already sent, which happens on almost every turn, so they make up most of the tokens in a long session. These are the rates per million tokens:
| Model | Input | Output | Cache write | Cache read |
|---|---|---|---|---|
| Claude Haiku 5.5 | $0.10 | $0.50 | $0.125 | $0.01 |
| Claude Sonnet 5.5 | $2 | $10 | $2.50 | $0.10 |
| Claude Opus 5.5 | $4 | $20 | $5 | $0.20 |
| Claude Fable 5.1 | $10 | $50 | $12.50 | $0.25 |
What Claude Code costs a developer on the API
Anthropic reports that across enterprise deployments, Claude Code averages about $13 per developer per active day and $150 to $250 per developer per month, and stays below $30 per active day for 90% of users. Your costs depend mostly on the model you choose, the size of the codebase, and how many sessions run at once.
Haiku 5.5 charges higher rates for prompts over 100K tokens. Batch processing costs 50% less, but Claude Code doesn’t use it for interactive work.
Limits
How Claude Code usage limits work
Subscription plans don’t publish a number of prompts or tokens. They set limits in two windows, and Claude Code stops when you reach either one:
- A five-hour session limit on Pro, Max, and Team, which resets on a rolling five-hour window.
- A weekly limit on every paid plan. Max and Ultra also have a separate weekly limit for Fable.
- The limits are shared across Claude chat, Claude Code, and cloud sessions. Running several sessions in parallel uses the allowance proportionately faster.
What happens when you reach a limit
You can wait for the reset, upgrade, or turn on usage credits. Usage credits keep Claude Code working past the limit and bill at standard API rates. You can set a monthly spend cap in Settings > Usage, and on Team and Enterprise an admin sets caps for the organization, groups, or individual members.
The monthly API credits included with Max and Ultra can’t be used in Claude Code or the Claude apps. If an ANTHROPIC_API_KEY environment variable is set, Claude Code uses that key instead of your subscription and bills the API account.
Why limits run out
Why Claude Code uses up limits and budget quickly
Anthropic’s documentation lists the main reasons a session uses more than the activity suggests. Each one comes from the fact that Claude Code re-sends the whole conversation with every request.
- Long sessions. A short question in a session that has been open all day still re-reads the full history.
- Cache misses. The first message after a break longer than the cache lifetime reprocesses the full context at the input rate.
- Subagents, agent teams, and workflows. Each one makes its own requests. Anthropic says agent teams in plan mode use about 7× the tokens of a standard session.
- Larger models and higher effort. Opus 5.5 costs twice as much as Sonnet 5.5 per token, and Fable 5.1 five times as much. Thinking tokens bill as output.
- Scheduled tasks and loops. They fire on their interval and send the full context each time, even while you are away.
Cost control
How to lower Claude Code costs
These changes make the largest difference, roughly in order. Run /usage in Claude Code to see which ones apply to you; on a subscription it attributes recent usage to skills, subagents, MCP servers, and loops.
- Use /clear between unrelated tasks, so each turn re-sends less history.
- Use Sonnet 5.5 for most coding, and save Opus and Fable for hard problems. Set simple subagents to Haiku.
- Lower the effort level with /effort for routine tasks.
- Keep CLAUDE.md short, and move workflow-specific instructions into skills, which load only when used.
- Turn off MCP servers you aren’t using, and prefer command-line tools such as gh.
- On the API, set workspace spend limits in the Claude Console. On Team and Enterprise, set spend limits for usage credits.
Compact the conversation earlier
Claude Code summarizes the conversation when it nears the auto-compact window. Setting that point lower makes it compact sooner, so fewer requests go out with a nearly full context, which are the most expensive ones. Set the CLAUDE_AUTOCOMPACT_PCT_OVERRIDE environment variable to a percentage below the default, for example export CLAUDE_AUTOCOMPACT_PCT_OVERRIDE=50; values above the default are ignored.
Cloud sessions
What it costs to run Claude Code in the cloud
Claude Code can run in the cloud so a task keeps going after you close your laptop. Anthropic’s own cloud sessions are included on Pro, Max, Ultra, and Team plans with no separate charge for the virtual machine; they count against the same usage limits. They clone from GitHub, and pull requests require GitHub.
Replicas, which publishes this guide, runs Claude Code in cloud workspaces using the Claude subscription or API key you already have. Replicas doesn’t resell model usage, so Claude usage stays on your Claude plan and Replicas charges for workspace time.
| Setup | Monthly price | Usage |
|---|---|---|
| Claude Code cloud sessions on a Claude plan | Included in Pro, Max, Ultra, and Team | Counts against the plan’s five-hour and weekly limits |
| Replicas Developer | $50 a month, including $50 of usage, plus your agent subscription | Every workspace counts, including API and automation runs, at $0.004 to $0.016 a workspace minute by size; usage beyond it bills at 1.5× |
| Replicas Team | $200+ a month in $100 steps, including the same amount of usage, plus your agent subscriptions | Shared by the whole team, at $0.004 to $0.016 a workspace minute by size; usage beyond it bills at 1.5× |
What you get in a Replicas workspace
Each task runs in its own Linux VM built from an environment your team defines once, with your dependencies, services, and a browser. Tasks start from the web, desktop, or iOS app, Slack, Linear, GitHub, GitLab, a schedule, or the API, and come back as a pull request. Replicas also runs Codex, OpenCode, Cursor, and Pi in the same workspaces.
When a Claude subscription reaches its limit, Replicas moves the chat along a fallback chain you set: another Claude credential, or Codex with the conversation carried over. This lets a team keep working instead of waiting for the reset or paying API rates.
Check Anthropic’s terms before using a personal subscription on a hosted platform. Anthropic’s legal and compliance page sets out how Claude subscription sign-in may be used. Use an API key for production pipelines.
Choosing a plan
Which Claude Code plan to choose
The right plan depends on how many hours a day you run Claude Code and whether you pay for yourself or a team.
| If you | Choose | Why |
|---|---|---|
| Use Claude Code a few times a week | Pro, or an API key | Pro is the cheapest flat price; an API key can cost less with light use |
| Use it every working day | Max | 5× Pro usage for $100 |
| Run several sessions at once or long agent runs | Ultra | 10× Pro weekly usage; verified accounts have no five-hour limit |
| Buy for a team of 2 to 150 | Team, mixing Standard and Premium seats | Central billing and SSO; give Premium seats only to heavy users |
| Need SCIM, audit logs, or usage-based billing | Enterprise, or the API through your cloud provider | Pay for what each developer uses |
FAQ
Claude Code pricing questions
Getting started with Replicas
Run Claude Code in the cloud on your own subscription
Replicas includes $100 of free usage with no credit card required. Connect a GitHub or GitLab repository, define the environment once, sign in Claude Code, Codex, OpenCode, or Cursor with the subscriptions or keys you already pay for, and start a task from the web, desktop, or iOS app, Slack, Linear, or a pull request comment. Each task runs in its own Linux VM and comes back as a pull request your team reviews.