SSD Nodes Learn Hosting plans →
How to do am Matt ConnorBy Matt Connor · Updated 2026-11-07

Claude Code dey included for Claude Pro?

Yes, Pro, Max, Ultra, Team and current Enterprise seats include Claude Code. Free no include am, while ANTHROPIC_API_KEY fit trigger per-token billing.

Claude Code dey included with Claude Pro?

Yes. Claude Code dey included for Claude Pro, Max, Ultra, every Team seat, and current Enterprise seat. So one subscription cover both the chat apps and the terminal tool. For older Enterprise contract, e depend on the seat type. Na only Premium seats and Chat + Claude Code seats include am, so check the seat wey you get. The free Claude plan no include am. Anthropic setup documentation talk am clearly: Claude Code need Pro, Max, Ultra, Team, Enterprise, or Console account. The free plan no give access to Claude Code. Subscription no need API key (application programming interface key, wey dem bill per token). Key na optional. If ANTHROPIC_API_KEY dey set and you approve am, dem go bill that session per token instead. This fit work for automated runs wey you want bill separately from your interactive work, or for one occasional session wey you choose to bill to specific key. Without subscription, Console account na the way, and dem go bill you per token.

The next part na where people dey make mistake. Claude Code no get separate allowance by itself. E dey use the same subscription usage budget wey your chat conversations dey use, so one long coding session fit reduce the chat window wey follow am. Anthropic talk say Pro, Max, and Ultra plans get usage limits wey Claude and Claude Code share. This mean say all activity for both tools count against the same usage limits. The only exception na current usage-based Enterprise plan, wey no get included budget to share. E no get per-seat usage limits, and dem go bill every request for either tool at API rates.

Wetin each plan really get

  • Free: e no get Claude Code. You fit chat for web and apps, but nothing dey your terminal.
  • Pro: Claude Code dey included for Pro usage levels. E cost $20 every month for United States as of October 2026, and annual billing fit cheaper.
  • Max: na the same Claude Code with five times Pro usage every five hours and every week. E cost $100 every month for web as of October 2026, and na monthly billing only. E include $100 API credits every month.
  • Ultra: na the same Claude Code with ten times Pro usage every week. E cost $200 every month for web as of October 2026, and na monthly billing only. E include $200 API credits every month. Mobile prices for Max and Ultra fit different.
  • Team and Enterprise: e dey included with every Team seat. The allowance belong to the seat, and premium seat get more allowance than standard seat. Current Enterprise plans use one seat type wey include Claude Code, with no per-seat allowance: dem bill all usage at API rates, under spend limits wey admin set. Older Enterprise contracts use different seat types: na only Premium seats and Chat + Claude Code seats include Claude Code, and na seat-based contracts get included allowance.

Na only these prices dey here, because plan prices dey change and this page no dey change with dem. Current numbers dey for full cost breakdown for Claude Code. You fit see how to choose between plans for plan by plan comparison of the Claude subscriptions, and wetin the free Claude tier really cover worth checking before you assume say you no get option.

Why one coding session dey shorten your chat window

Two windows dey control Pro, Max or Team subscription, plus seat for older seat-based Enterprise contract. Session limit dey reset every five hours. Weekly limit dey reset for fixed time every week wey dem assign to your account, and you fit see your own reset time for usage screen. Verified Ultra accounts no get five-hour limit; unverified accounts get temporary limit. Ultra still get weekly limit across all models, and Max and Ultra cap Fable at half of the regular weekly allowance. E dey use that same budget and finish am faster than other models. Current usage-based Enterprise plan no get either window: e bill consumption, and admins cap am with spend limits instead. Neither window be message count wey you fit calculate ahead of time, because how far you go depend on how long your conversations be and which model you run. Treat dem as behaviour to monitor, no be number to budget against. The limit mechanics in detail dey worth reading once before you hit one.

Claude Code dey drain that budget faster than chat because of wetin coding turn carry. Every request send the whole conversation wey don happen so far. Each tool call, file read, or test run send another request wey carry the result from the previous one. Anthropic own guidance for administrators talk say make you budget more for coding seat than chat seat for exactly this reason: one debugging session fit consume more than one day of chat.

Prompt caching dey reduce this effect, but e no remove am. Repeated context dey read again at cached rate, so one-line question inside session wey don stay open all day still dey use resources for the entire conversation behind am. For subscription, while you still dey within the usage wey your plan include, the main conversation get one-hour cache lifetime. Subagents, compaction, and most other requests wey Claude Code make outside that conversation default to five minutes. If you step away pass one hour, your first message when you come back no go find the cache, so your full context go write to the cache again at the cache-write rate. That rate pass normal input rate; e no be the same: Anthropic price one-hour cache writes at 2 times the base input price, and five-minute cache writes at 1.25 times. That one message fit cost more than the twenty messages before am.

Signs wey show say you dey use the shared budget finish

Run /usage inside Claude Code. If you dey use paid plan, e go show usage bars for your plan limits, plus breakdown of wetin dey use dem. Press d for the last 24 hours and w for the last 7 days.

The breakdown go flag any behaviour wey cause 10 percent or more of recent usage, like long context or cache misses. E go also show how skills, subagents, plugins, and individual MCP (model context protocol) servers use the budget. One important thing be say these figures come from local session history for the machine wey you dey use. So work wey you do for another machine or for claude.ai no dey inside the breakdown. The plan bars dey apply globally. The attribution dey apply only locally.

When you finish the available usage, the message go name the ceiling wey you reach:

You've hit your session limit
You've hit your weekly limit

Both windows dey apply across every model, so if you switch model with /model, e no go restore your access. A model-specific message like You've hit your Opus limit na different matter, and for that case /model go let you continue work with another model. Context warning or auto-compact notice no be usage limit at all. E mean say the conversation dey near the session context window, and older history go soon become summary. Anthropic cost guidance put context management first, and keeping the context window small get the clearest effect on how fast the budget dey finish. /clear between unrelated tasks no cost anything, and e start the next request with less old context to carry. If dem don lock you out already, the options when a limit is reached na to wait, upgrade, or turn on usage credits. If Anthropic give your account free limit reset, that one na fourth option. Use am from Settings > Usage for the web or inside Claude Desktop, not from Claude Code itself. The session or weekly limit wey e cover go return to full for Claude Code too.

Wetin make API key fit take over your bill

Dis one dey cause serious wahala for server. If ANTHROPIC_API_KEY environment variable dey set and you approve am, Claude Code go authenticate with dat key instead of your subscription, and e go bill the work per token to your API account. For interactive session, e go ask once whether e fit use the key and remember your answer. So, key wey you approve months ago fit continue to take over without another prompt. To change dat choice later, use the "Use custom API key" toggle for /config. E go show only while the variable dey set. For non-interactive mode (claude -p), e go use any key wey dey present, without prompt at all. Your Pro, Max or Ultra allowance go remain untouched while the invoice dey grow. No free allowance dey wait for the other side too, because Claude API no get free tier apart from the small credit wey dem grant when you sign up. The monthly API credits wey come with Max, Ultra and Team plans no cover interactive Claude Code. Dem cover claude -p runs and Agent SDK work wey you start yourself with API key from the Console organization wey link to your plan.

For VPS, dis one dey happen by mistake more often than for laptop, because the same shell profile usually dey hold keys for other tools. Anthropic own advice na make ANTHROPIC_API_KEY remain unset if you want your subscription to pay. Set key only for the terminal session wey need am. After you log in, Claude Code also show notice when subscription login and API key both dey active. Check before you start:

printenv ANTHROPIC_API_KEY

If nothing print, e mean no API key dey set for dis shell. But dis alone no prove say Claude Code go use your subscription. Credential fit also come from ANTHROPIC_AUTH_TOKEN, an apiKeyHelper script, or env block inside settings file, including settings wey administrator dey manage. Any of dem get priority over claude.ai login. Run /status inside Claude Code to see which one dey active. If key print, the next session go bill the API if you approve dat key before or run claude -p. Otherwise Claude Code go ask first, and if you decline, e go keep you on the subscription. Clear am for dis shell, then find the file wey dey set am again:

unset ANTHROPIC_API_KEY
grep -rn ANTHROPIC_API_KEY ~/.bashrc ~/.profile ~/.bash_profile 2>/dev/null

Now start Claude Code and sign in with the subscription account:

claude

The first run go start browser login. Inside session, /login dey switch accounts, /status show the authentication method wey dey use, and /usage confirm which billing account you enter: subscriber go see plan usage bars, while API account go see session block with token costs and no plan bars. The remaining server-side setup, including how to keep session alive after you disconnect, dey covered for how to run Claude Code for VPS inside tmux. The permission side dey covered for how to run Claude Code safely for server wey belong to you.

One related trap dey. When plan allowance finish, you fit turn on usage credits to continue work beyond am, with extra billing on top of subscription. Once you start use credits, Claude Code reduce the main conversation prompt cache lifetime from one hour to five minutes. So, the same idle gaps wey dey free before go start cost you, unless you set promptCacheTtl setting to 1h yourself. /usage-credits open Settings > Usage for Pro, Max or Ultra subscriber.

Which one you suppose pay for?

All this one assume say you don decide to pay at all, and whether Claude Code go earn back the money for your own hourly rate na the arithmetic behind that decision. If wetin you dey consider na the subscription itself instead of the terminal tool, whether Claude Pro worth am for how you actually dey use Claude start from another question: how often the free limit dey stop you in the first place. If you don already dey pay and you dey think to stop, wetin you go keep and wetin you go lose if you cancel Claude Pro show say your chats and Projects go still dey, but Claude Code go stop when your paid time finish.

  • Occasional or hobby coding: Pro. One tool, one bill, enough for some sessions every week. Details of the allowance dey for the Pro plan price and its limits.
  • Sustained daily coding: Max or Ultra. Max gives five times Pro usage for both windows; Ultra gives ten times Pro usage every week and removes the five-hour limit for verified accounts. The choice between Max and Ultra depend on how often you dey reach the weekly limit and whether you need work without five-hour ceiling.
  • Several people: Team or Enterprise seats, where the allowance follow the seat (or, for current usage-based Enterprise plan, every request dey billed at API rates under admin spend limits) and admin fit see how much each user spend. The cost of one seat compared with the cost of the same work per token na the comparison to run before you buy seats in bulk.
  • Bursty or automated runs: API billing. Nightly job or CI (continuous integration) step no get reason to compete with your interactive work for the same allowance.

Those last two cases fit work together well. Keep the subscription for the work wey you do by hand, and give automation its own API key, so overnight run no go empty the window wey you need the next morning. The API and subscription cost comparison dey explain where the crossover dey.

FAQ

I fit use Claude Code for the free Claude plan?

No. Claude Code need Pro, Max, Ultra, Team, Enterprise or Console account. Anthropic setup documentation talk say free Claude plan no include Claude Code access, so when you run claude, e go ask you to sign in to paid account or authenticate with API key instead. Free plan still cover chat for web and desktop apps. That one enough to test whether Claude fit answer your kind question before you pay for anything.

Claude Code dey use the same usage limits like Claude chat?

Yes for Pro, Max and Ultra. Anthropic talk say Claude and Claude Code share usage limits, so all activity for both tools dey count against the same limits. Pro and Max get session limit wey reset every five hours, and all three plans get weekly limit wey reset for fixed time every week assigned to your account. Verified Ultra accounts no get five-hour limit; unverified accounts get temporary limit. Max and Ultra limit Fable to half of the regular weekly allowance. E dey use that same budget and finish am faster than other models. For real use, one afternoon of heavy refactoring fit leave you with short chat window that evening. Run /usage to see the bars and the next reset time.

Why my usage drop so fast during one coding session?

Na because coding turn carry much more context than chat message. Every request send the whole conversation, and each tool result add another request wey carry the conversation again. Run /usage and read the behaviour flags. Anything wey cause 10 percent or more of recent usage go get mark, and the usual causes na long context and cache misses. Session wey stay open all day go keep sending the full history again. Idle gap wey pass the cache lifetime (one hour for the main conversation on a subscription within plan usage, five minutes for subagents and once you dey use usage credits) make the next message write everything to cache again at the cache-write rate. This rate pass normal input rate: 2 times base input for one-hour cache write and 1.25 times for five-minute one. Using /clear between unrelated tasks address both problems.

When API key cheaper pass Pro or Max subscription?

When the work dey come in bursts or na automation dey run am. API billing charge per token with no monthly minimum, so script wey run twice every week cost only wetin those tokens cost and nothing more. Subscription dey pay for one month capacity wey you no use. Daily coding wey dey run continuously work the other way: subscription na fixed price for large allowance, while paying per token for the same volume cost more. Setting ANTHROPIC_API_KEY for machine wey you also dey use to code by hand na the mistake to avoid. Once you approve the key for the prompt, or run claude -p while the key dey present, Claude Code go bill the API and leave your subscription allowance unused.

Dis article help you?