Claude Code Pricing (October 2026): Pro vs Max, Usage Limits and What It Really Costs

Claude Code pricing at a glance

Claude Code pricing comes down to one choice: pay a flat monthly fee through a Claude subscription, or pay per token through the API. As of October 10, 2026, Claude Code is included in Claude Pro at $20 a month ($17 a month if you pay $200 up front for a year), Max 5x at $100 a month and Max 20x at $200 a month. The free Claude plan does not include it.

Teams pay per seat. A Team Standard seat costs $25 a month ($20 billed annually), and a Team Premium seat costs $125 a month ($100 annually) with five times the usage of a Standard seat. Enterprise starts at $20 per seat per month, billed annually, plus usage at API rates. Every paid tier includes Claude Code, according to Anthropic's pricing page.

The sticker price is the easy part. What the price doesn't tell you is how much coding you get before Claude stops and tells you to wait. That's the question most people searching for Claude Code pricing are really asking, so most of this guide is about limits.

• Free: $0, no Claude Code.
• Pro: $20 a month, or $17 a month billed annually. Claude Code included.
• Max 5x: $100 a month, 5x Pro's per-session usage.
• Max 20x: $200 a month, 20x Pro's per-session usage.
• API: pay per million tokens, from $0.10 (Haiku 5.5 input) to $50 (Fable 5.1 output).

Is Claude Code free?

No. The Claude Code app itself is free to install, but it won't do anything until you sign in with a paid Claude plan or an API account with credits. The free Claude plan covers chat on the web, desktop and mobile, not Claude Code.

There are two partial workarounds. You can point Claude Code at other model providers, which we walk through in how to get Claude Code free with OpenRouter. You give up Anthropic's models when you do. Or you can use a cheaper coding plan from another lab, such as the GLM Coding Plan.

One trap worth knowing: the Max plans now come with $100 to $200 a month in API credits for building on the Claude Platform. Those credits can't be spent in Claude Code or the Claude apps, according to Anthropic's help center. They're for your own API projects.

Claude usage limits: what Pro and Max really get you

Every Claude plan runs on two meters. A five-hour session limit resets five hours after you start, and a weekly limit resets at a fixed time shown in Settings > Usage. Claude chat, Claude Code and IDE use all draw from the same pool, so a long chat in the browser eats into your coding time.

Anthropic doesn't publish message or token counts for any plan. It describes limits only as multiples: Pro gets at least 5x the free plan per session, Max 5x gets 5x Pro, and Max 20x gets 20x Pro. How far that goes depends on conversation length, model choice, tool use and effort level. Max also has a separate weekly limit for Fable models. On Pro, Fable is available only through paid usage credits.

Limits also move. Anthropic ran a temporary 50% boost to weekly Claude Code limits through mid-September, then set a permanent 25% increase over the old baseline from September 14, 2026. In its own clarification, Anthropic said that works out to "a 17% reduction in weekly limits" compared with the promotion, BleepingComputer reported.

When you hit a limit, you have four options: wait for the reset, turn on usage credits (billed at API rates), upgrade, or buy API credits in the Claude Console. Run /usage inside Claude Code to see your plan bars and which skills, subagents and MCP servers are using the most.

Claude Pro vs Max: which plan should you pick?

Divide each plan's price by the usage it buys and a pattern shows up. Pro costs $20 for 1x. Max 5x costs $100 for 5x, which is still $20 per unit: you're buying more, not buying cheaper. Max 20x costs $200 for 20x, which works out to $10 per unit, half the price. It's the only tier with a real volume discount.

That makes the choice simpler than the Claude Max price suggests. If Claude Pro limits stop you more days than not, more usage is worth paying for. If they don't, Pro is enough. The Claude Max plan is billed monthly only, and upgrades are prorated for the rest of your billing cycle, so you can move up mid-month without losing money.

• Pick Pro if you code with Claude a few hours a week and can live with an occasional five-hour wait.
• Pick Max 5x if you hit Pro's session limit most working days. It's the same rate per unit, just more of it.
• Pick Max 20x if you run long agent sessions, parallel subagents or agent teams, or you still hit weekly caps on Max 5x. It's the only plan that lowers your cost per unit.
• Before any upgrade, cut wasted tokens first (see below). A leaner setup can make a $20 plan feel like a bigger one.

Claude Code API pricing: when pay-per-token wins

If you sign in with an API key, you pay per million tokens instead of a flat fee. Sonnet 5.5 costs $2 per million input tokens and $10 per million output tokens. Opus 5.5 costs $4 and $20, Fable 5.1 costs $10 and $50, and Haiku 5.5 costs $0.10 and $0.50 for prompts up to 100K tokens. Cache reads are far cheaper, which matters because Claude Code re-sends your conversation on every turn.

So what does Claude Code cost on the API in practice? Anthropic's own docs put the average at about $13 per developer per active day and $150 to $250 per developer per month across enterprise deployments, with 90% of users staying under $30 a day.

That puts a typical daily API user above Max 5x and close to Max 20x. For anyone coding most days, a subscription is the cheaper way to pay. The API makes sense when you use Claude Code a few days a month, run it in CI or other automation, or need per-user spend limits for a team.

Watch for one billing surprise: if an ANTHROPIC_API_KEY environment variable is set on your machine, it overrides your subscription sign-in and Claude Code bills the API instead. Run /status to check which account a session is using.

Claude Code vs Codex pricing

OpenAI prices Codex the same way: inside a ChatGPT plan. ChatGPT Plus is $20 a month and includes Codex with five-hour limits. ChatGPT Pro starts at $100, with $200 and $500 tiers, and OpenAI says Pro plans currently have no five-hour limit. The $8 Go plan and the free plan are limited to GPT-6 Luna in the desktop app.

Price alone won't settle it, because both companies count usage differently and change their limits often. If you're deciding between the two tools, our Claude Code vs Codex comparison covers the quality side, and the Codex usage limits guide covers OpenAI's meters.

How to stop hitting the limit you already pay for

Before you move from $20 to $100, look at where your tokens go. Anthropic's cost docs list the usual causes, and most of them are settings, not habits you have to remember:

• Stale context: Claude Code re-sends the whole conversation on every request, so a session left open all day costs more per message. Use /clear between unrelated tasks.
• A long CLAUDE.md: it loads into every session. Anthropic suggests keeping it under 200 lines and moving specialized instructions into skills.
• The wrong model: Sonnet handles most coding tasks for half of Opus's API price. Save Opus and Fable for hard problems.
• Noisy tool output: a 10,000-line test log in context burns tokens. A hook that keeps only failures cuts it to a few hundred.
• Extra MCP servers and high effort settings you don't need for simple tasks.

The Token Saver Kit: the fixes, already written

We packaged those fixes into the Token Saver Kit for Claude Code and Codex. It's the setup we use, as files you copy into a project: a lean CLAUDE.md and AGENTS.md template, a settings.json that blocks reads of dependencies, build output, lockfiles and .env files, two hooks (one stops whole-file reads of huge files, one trims test output to failures), two cheap Haiku subagents that return short summaries, a codebase-map skill, and Codex config defaults.

Setup takes about 10 minutes, and every file comes with a note explaining why it works, with links to Anthropic's and OpenAI's docs. It costs $24 once. We won't promise a specific saving, because it depends on how you work. The guide shows how to measure before and after with /usage and /context.

If you'd rather do it by hand, our free guides cover the biggest wins: saving tokens in Claude Code with Graphify and saving tokens in Codex.

Claude Code pricing FAQ

How much is Claude Code per month? $20 on Pro ($17 billed annually), $100 on Max 5x, $200 on Max 20x, or $25 per Team Standard seat. On the API there's no monthly fee; Anthropic's average is $150 to $250 per developer per month.

Is Claude Code included with Claude Pro? Yes. Pro, Max, Team and Enterprise all include Claude Code. The free plan doesn't.

Do Claude Code and Claude chat share usage limits? Yes. Claude, Claude Code and IDE use count toward the same five-hour and weekly limits.

When do Claude usage limits reset? The session limit resets five hours after a session starts. The weekly limit resets at a fixed time assigned to your account, shown in Settings > Usage.

Is Claude Max worth it? Max 20x is the only plan that lowers your cost per unit of usage. Max 5x is worth it only if you hit Pro's limits most days.

Can I use the API credits that come with Max in Claude Code? No. Anthropic says those credits are for building on the Claude Platform, not for Claude Code or the Claude apps.

What to remember

• Claude Code costs $20 (Pro), $100 (Max 5x) or $200 (Max 20x) a month, or pay-per-token on the API. The free plan doesn't include it.
• Limits run on a five-hour window and a weekly cap, shared with Claude chat. Anthropic doesn't publish exact numbers.
• Max 20x is the only plan that halves your cost per unit of usage.
• Typical API use runs $150 to $250 a month, so daily users usually pay less on a subscription.
• Cut wasted tokens before you upgrade.

Sources