Pricing guide
Claude Code Pricing (2026): Claude Code Cost Per Month, Max Plans and Team Seats
Every plan that includes Claude Code, what it costs, and the usage windows that decide which tier you actually need. All figures read from Anthropic's own pricing page and cost documentation in July 2026, with the flat-price way to run an agent on your repo alongside.
No card to start · Review-first: the agent never merges
Pick a task
Plan
- planning
Files changed
Test run
Pull request
You review and merge. Agentcode never merges on its own.
The short answer
Claude Code pricing in July 2026 is bundled into the Claude subscription rather than sold separately: Free ($0, smallest allowance), Pro ($20 a month, about $17 annual), Max 5x ($100 a month), Max 20x ($200 a month), and Team seats from about $20 each annually with a $100 premium seat for heavy users. Paid plans do not bill per token; usage draws from an allowance that resets on rolling five-hour and weekly windows, so the tier you need depends on how hard you run the agent. Anthropic's own documentation puts real token-billed usage at roughly $13 per developer per active day and $150 to $250 per developer per month. The flat-price way to run an agent on your repo, with no usage window at all, is Agentcode at $29 a month.
Claude Code plans and prices
Claude Code is not sold on its own. It ships with the Claude subscription, so the price you pay is the price of a Claude plan. The column that decides your real cost is the allowance, because that is what an agent session draws down before it stops.
| Plan | Price | Usage allowance | Best for |
|---|---|---|---|
| Free | $0 free | Smallest allowance | Trying Claude Code out. |
| Pro | $20 per month | Baseline paid allowance | One developer, moderate use. |
| Max 5x | $100 per month | About 5x Pro | A developer coding with it daily. |
| Max 20x | $200 per month | About 20x Pro | Heavy, all-day agentic work. |
| Team (Standard seat) | $25 per seat / mo | More than Pro | A small team wanting admin control. |
| Team (Premium seat) | $125 per seat / mo | About 5x a standard seat | The heaviest members of a team. |
| Enterprise | Custom contact sales | Seat allowance plus credits | Larger orgs with procurement needs. |
| Console (API) | Per token pay as you go | Uncapped until you cap it | Automation and CI usage. |
Prices as published by Anthropic and checked in July 2026. Annual billing lowers the monthly figure on Pro and Team seats. Anthropic revises plans and usage windows regularly, so confirm on claude.com/pricing before you buy.
What each Claude Code plan gets you
Free
$0 · free · Smallest allowanceClaude Code is included on the free plan with the smallest usage allowance of any tier. Enough to run it on a small repo and see how the agent reasons about your code, not enough to lean on it through a working day. Most developers hit the session limit quickly and move to Pro.
Pro
$20 · per month · Baseline paid allowanceThe entry paid plan and the one most individual developers start on. It drops to about $17 a month on annual billing. Pro covers Claude chat, Claude Code and Cowork out of one shared allowance, which matters more than it sounds: heavy chat use eats into the budget you have left for coding sessions.
Max 5x
$100 · per month · About 5x ProRoughly five times the Pro allowance. This is the tier most full-time Claude Code users land on, because agentic sessions consume far more than chat: every turn carries file contents, tool calls and multi-step reasoning. If you are hitting the five-hour limit on Pro most afternoons, this is the fix.
Max 20x
$200 · per month · About 20x ProAbout twenty times the Pro allowance, aimed at developers running long agent sessions or several instances at once. Max plans carry a weekly cap layered on top of the rolling five-hour window, so even at $200 the usage is bounded rather than unlimited.
Team (Standard seat)
$25 · per seat / mo · More than ProAbout $20 per seat on annual billing, $25 billed monthly, for teams of roughly 2 to 150 users. Claude Code is included on the standard seat, and each member draws from a per-seat allowance shared across Claude chat, Claude Code and Cowork. Adds central billing, admin settings and org analytics.
Team (Premium seat)
$125 · per seat / mo · About 5x a standard seatAbout $100 per seat annual, $125 monthly, with roughly five times a standard seat's allowance. Seats can be mixed, so a team typically buys premium seats for the few engineers who live in the agent and standard seats for everyone else.
Enterprise
Custom · contact sales · Seat allowance plus creditsSeat-based pricing with usage-based costs on top, plus SSO, audit logs, the Enterprise Analytics API for per-user cost reporting, and spend limits set at the org, group or member level. This is the tier where finance can actually see and cap what Claude Code is costing per engineer.
Console (API)
Per token · pay as you go · Uncapped until you cap itAuthenticate Claude Code against the Claude Console and it bills per token instead of drawing on a subscription: Sonnet at $3 in and $15 out per million tokens, Opus at $5 and $25, Haiku at $1 and $5. No usage windows to hit, but no ceiling either until you set a workspace spend limit.
What Claude Code actually costs per developer
The sticker price and the budget line are different numbers, and Anthropic is unusually open about the gap. Its own cost documentation publishes what real deployments spend, which makes planning a rollout much easier than guessing from the plan page.
| Measure | Published figure | What it means for a budget |
|---|---|---|
| Average per developer, per active day | About $13 | An active day means a day someone actually ran it, not every working day. |
| Average per developer, per month | $150 to $250 | This, not $20, is the number to take to finance for a token-billed team. |
| 90th percentile, per active day | Below $30 | Nine in ten developers stay under $30 a day, so the tail is real but narrow. |
| Agent teams overhead | About 7x a standard session | Running multiple teammates multiplies context, and therefore spend. |
Figures published by Anthropic in its Claude Code cost documentation, read in July 2026. They describe token-billed deployments; subscription seats are capped by usage windows instead.
Why coding costs more than chat on the same plan
On a Claude plan your allowance is shared across chat, Claude Code and Cowork, and coding is by far the heaviest of the three. Every agent turn carries file contents, tool call results and multi-step reasoning, so one debugging session can consume more of your budget than a full day of conversation. That is the single biggest reason developers who were comfortable on Pro for chat find themselves upgrading within a week of adopting the agent.
Long sessions cost more than the work in them
Claude Code resends the conversation with every message, so a one-line question asked in a session that has been open all day is priced against the whole conversation, not the one line. Cache misses after a break, scheduled tasks firing on an idle session, and compaction passes all add to that. Clearing between unrelated tasks is the cheapest habit a team can adopt, and it routinely matters more than which model you picked.
Two limits, not one
Subscription plans meter on a rolling five-hour window and a weekly window layered on top. Both apply across all models, which is why switching from Opus to Sonnet with the model command does not restore access once you have hit a session or weekly limit. Teams that need to work past the allowance turn on usage credits, and that overflow bills at API rates, so the predictable plan quietly becomes a variable one.
Which tier you need, honestly
Try Pro first and watch for one signal: whether you hit the five-hour limit in the middle of real work. If it happens once a fortnight, stay on Pro. If it happens most afternoons, Max 5x at $100 is the honest cost of your workflow, and paying it beats losing an hour waiting for a reset. Max 20x at $200 only pays for itself if you run long sessions or several instances at once. For a team, buy premium seats for the two or three engineers who live in the agent and standard seats for everyone else rather than upgrading the whole org.
Claude Code vs a flat-priced PR agent
Claude Code is an interactive agentic CLI you drive from your terminal; Agentcode is an async agent that takes a task and hands back a pull request. They solve different halves of the same problem, so here is where each one is genuinely ahead.
| Claude Code | Agentcode | |
|---|---|---|
| Billing model | Subscription with rolling five-hour and weekly usage windows, or per-token API billing | Flat subscription, no usage window and no per-token meter |
| Entry price | $20 / mo (Pro), about $17 annual | From $29 / mo, flat |
| What a busy month costs | More: heavy days hit session and weekly limits, so you upgrade a tier or buy usage credits | The same as a quiet month |
| Where it runs | Your terminal, IDE extensions and CI, on any local repository | Hosted, connected to your GitHub or GitLab repo |
| Depth of codebase reasoning | Excellent: subagents, skills, hooks and MCP make it the strongest agentic CLI available | Narrower by design: one task in, one reviewable pull request out |
| Supervision needed | Interactive: you drive the session and approve as it works | Async: you write the task, then read the PR when it lands |
| Opens pull requests | Yes, through the GitHub app or by running git in the session | Yes, on GitHub and GitLab, as the whole point of the product |
| Runs your test suite before the PR | Yes, when you tell it to in the session | Yes, every run, with results reported in the PR |
| Trains on your code | No | Never, on any plan |
Claude Code wins clearly on depth and control. If you want to sit with an agent, watch it reason about your codebase and steer it turn by turn, nothing in the category does it better right now, and the plan price is fair for what you get. Where a flat plan wins is predictability and delegation: Agentcode runs the same async PR workflow on GitHub and GitLab for one monthly price, so a busy sprint and a quiet week cost the same and nobody stops mid-task waiting for a window to reset. Plenty of teams run both, interactive work in one and queued background tasks in the other. For the fuller breakdown see the Claude Code alternative page and the flat vs metered pricing comparison across every tool.
Claude Code pricing: your questions answered
How much does Claude Code cost?
Claude Code is included in every paid Claude plan. As of July 2026 that means $0 on Free with a small allowance, $20 a month on Pro (about $17 annual), $100 a month on Max 5x, $200 a month on Max 20x, and Team seats from about $20 each annually. You can also run it against the Claude Console and pay per token instead of subscribing.
Is Claude Code free?
Claude Code is available on the free Claude plan, but the usage allowance is the smallest of any tier and agentic sessions burn through it fast. It is enough to try the tool on a small repository and judge the quality of its reasoning. Sustained daily coding work realistically starts at the $20 Pro plan, and most full-time users end up on Max.
Do I need Claude Max for Claude Code?
Not necessarily, but many daily users end up there. Pro at $20 covers moderate use, and the signal that you have outgrown it is hitting the five-hour session limit most afternoons. Max 5x at $100 gives roughly five times the allowance and Max 20x at $200 about twenty times. Coding sessions consume far more than chat because every turn carries file contents and tool calls.
What does Claude Code actually cost per developer?
Anthropic publishes real numbers for this. Across enterprise deployments the average is around $13 per developer per active day and $150 to $250 per developer per month, with 90 percent of users staying below $30 per active day. Those figures are for token-billed setups, and they are the honest planning baseline for a team, not the $20 sticker price.
Does Claude Code charge per token?
It depends how you authenticate. On Pro, Max, Team and Enterprise subscriptions you do not pay per token: usage draws from a seat allowance that resets on rolling five-hour and weekly windows. On the Claude Console or a cloud provider like Bedrock, Claude Code bills per token at API rates, which is why teams automating it in CI cap spend with workspace limits.
What happens when you hit the Claude Code limit?
The session stops and tells you when the window resets. Switching models does not restore access, because the session and weekly windows apply across all models. If your organization has usage credits turned on, you can keep working past the allowance and the extra usage bills at API rates. Otherwise you wait for the reset or move up a tier.
Is Claude Code worth it?
For interactive agentic coding it is the strongest tool in the category, and $20 is a fair price to find that out. The thing to model before you standardize a team on it is not the subscription but the usage ceiling, since that is what pushes engineers from $20 to $100 to $200. Budget against Anthropic's own $150 to $250 per developer per month figure rather than the entry price.
Is there a flat-price alternative to Claude Code?
Yes. Agentcode charges a flat fee from $29 a month with no usage window and no per-token meter: you describe a task, it plans, edits the code, runs your test suite and opens a reviewable pull request on GitHub or GitLab, for the same price whether you run it once or fifty times. It is async rather than interactive, so plenty of teams run both.
Last updated: July 2026
Weighing Claude Code against the field?
These pages compare pricing and reach tool by tool: Claude Code alternative, Cursor pricing, GitHub Copilot pricing and the Agentcode vs Claude Code comparison.
Or step back to AI coding tool pricing across every tool, the best AI for coding in 2026, or what an AI coding agent does before you weigh the bill.
A flat bill, on your repo
Connect a GitHub or GitLab repository, describe a task, and read the pull request it opens. Same price whether you run it once or a hundred times, with no usage window to wait out.