Claude Code itself costs nothing: you pay for the model working behind it, and you can do so in two ways. With a subscription, Claude Code is included in the plan: Claude Pro costs $20 per month ($17 per month with the annual subscription, $200 paid upfront), Claude Max costs $100 or $200 per month. These are Anthropic's dollar prices, excluding taxes; the free plan does not include Claude Code, and a subscription purchased on the website can be paid for only by card. With an API key, there is no subscription fee and you pay based on usage: on Kunavo, Claude Sonnet 5 costs $1,40 per million input tokens and $7,00 per million output tokens, so a typical agent step (approximately 25.000 input tokens and 1.200 output tokens) costs about $0,0434. The Kunavo balance can be topped up from $10, including with Satispay, which is offered at checkout in euros.
To understand which is better for you, convert the subscription fee into steps: with the Kunavo key, Pro's $20 buys approximately 460 steps of this type. Do you do more each month and do Pro's limits not stop you? Then the subscription is the better deal. Do you program in phases, perhaps only on weekends or when a project is due? The key costs less, because in a month when you do not use it, you pay nothing.
Current as of October 3, 2026. I prezzi dei piani vengono dalla Claude pricing page in Italian e dal centro assistenza di Anthropic: Pro plan, Max plan e Payment FAQs. I prezzi Kunavo vengono dal catalogo dei modelli (aggiornati al October 3, 2026). Gli importi sono in dollari perché in dollari li pubblicano sia Anthropic sia Kunavo.
How much does Claude Code cost? Subscription or API key
| Subscription (Pro / Max) | API key (usage-based) | |
|---|---|---|
| What you pay | A fixed monthly fee of $20 (US price, excluding taxes) | The tokens you use, with no subscription fee |
| Limits | A session limit that resets every five hours plus a weekly limit, shared with Claude chat | No usage window; the balance is the cap |
| Quiet month | Full subscription fee | $0 |
| Intensive month | Wait for the reset, or use usage credits at API rates | You keep working; the cost grows with usage |
| How you pay | On the website, by card only; in the app, through the App Store or Google Play | On Kunavo: Satispay, card, Link, Apple Pay, Google Pay |
| Cost control | Usage bars in /usage | Per token, for each key |
| Who it is for | Daily, consistent work by one person | Periodic work, per-person budgets, CI, and automations |
Le due strade si escludono a vicenda. Se imposti la variabile ANTHROPIC_AUTH_TOKEN (oppure ANTHROPIC_API_KEY, che in modalità interattiva va approvata una volta), Claude Code usa la chiave al posto dell'abbonamento: i limiti del piano smettono di valere e i token li paga il titolare della chiave (LLM gateway documentation e Article on Claude Code with Pro and Max plans, letti il October 3, 2026). Il login di claude.ai resta salvato ma non viene usato; togli la variabile e Claude Code torna all'abbonamento (Documentation on connecting to a gateway). Quale dei due è attivo lo mostra il comando /status.
Claude Code by subscription: Pro, Max, and Team
Secondo la Installation documentation, Claude Code richiede un account Pro, Max, Team, Enterprise o Console; il piano gratuito di claude.ai non lo include. L'Italia è nell'List of countries in cui Anthropic offre sia l'API sia Claude.ai. Prezzi da claude.com/it/pricing e dall'articolo sul piano Max, in dollari, imposte escluse (letti il October 3, 2026):
| Plan | Price (USD, excluding taxes) | Usage |
|---|---|---|
| Free | $0 | Without Claude Code |
| Pro | $20 per month; $17 per month with the annual subscription ($200 upfront) | Base limit, Claude Code included |
| Max 5x | $100 per month | Five times Pro usage |
| Max 20x | $200 per month | Twenty times Pro usage |
| Team, standard seat | $20 per seat per month with annual payment, $25 with monthly payment | Team plan |
Secondo Anthropic i prezzi di Max valgono per gli abbonamenti acquistati sul web; nelle app per smartphone possono essere diversi. Secondo l'Article on Fable models (letto il October 3, 2026), con Pro Fable 5 e Fable 5.1 non rientrano nei limiti del piano e si usano con i crediti di utilizzo (a pagamento, a tariffe API), mentre con Max sono inclusi fino al 50% dei limiti settimanali.
Price in euros and VAT
Anthropic indica i prezzi in dollari, e la sua pagina prezzi in italiano precisa che non includono le imposte applicabili. Secondo l'Article in Italian on the Pro plan, Pro costa $20 al mese negli Stati Uniti, con prezzi nella valuta locale dove supportata; i prezzi mensili variano a seconda della regione, e in alcune regioni l'imposta è già inclusa nel prezzo mostrato, in altre si aggiunge al momento del pagamento. L'aliquota IVA ordinaria in Italia è del 22% (Your Europe, European Commission, letto il October 3, 2026). Secondo Anthropic la giurisdizione fiscale la determina l'indirizzo di fatturazione e, a seconda del paese, la partita IVA si può inserire nelle impostazioni dell'account, in Settings > Billing (Article on the VAT identification number, letto il October 3, 2026).
The euro figures found in Italian guides do not match because each calculates them differently: dollars converted at the day's exchange rate plus VAT, or the App Store price. A subscription purchased in the iPhone or Android app is charged by Apple or Google and may cost differently from the website. The amount that matters is what Anthropic displays at payment; that is why we do not state a euro price here.
Satispay or PayPal for Claude Pro on the website? No
Secondo le Anthropic payment FAQs, gli abbonamenti acquistati sul web si pagano solo con carta di credito o di debito, e per gli acquisti in app il pagamento lo gestiscono l'App Store o Google Play, con i metodi di pagamento del tuo account Apple o Google. Anthropic non accetta nemmeno processori di pagamento di terze parti come PayPal (article about declined cards, letto il October 3, 2026). Satispay serve solo per la strada API, tramite Kunavo: lo spiega la sezione sul pagamento più sotto.
Limits: what you actually pay with the subscription
Pro e Max misurano l'utilizzo con un limite di sessione che si azzera ogni cinque ore e con un limite settimanale valido per tutti i modelli (articolo sul piano Pro). Entrambi sono condivisi tra la chat di Claude e Claude Code: tutto quello che fai nell'una e nell'altro pesa sullo stesso limite (articolo su Claude Code con i piani Pro e Max). Anthropic non pubblica numeri precisi; come funzionano le finestre di utilizzo lo spiega in dettaglio la guida Claude limits (in inglese). Secondo la Cost documentation, i limiti di sessione e settimanali valgono per tutti i modelli insieme, quindi cambiare modello con /model non li aggira; aiuta solo quando hai esaurito il limite specifico di un modello.
Quando il limite finisce puoi aspettare il reset, passare a un piano superiore o attivare i crediti di utilizzo. Sono prepagati e addebitati alle tariffe API standard (Anthropic help center), e la cache in quel caso dura di default cinque minuti invece di un'ora (Cost documentation, entrambi letti il October 3, 2026). So beyond the limit, you still pay the API price – ed è qui che serve una chiave tua: su Kunavo i modelli Claude costano il 30% in meno rispetto al listino Anthropic.
Is Claude Code free?
The tool is free; using it is not. La CLI e le estensioni per gli editor non costano nulla; paghi le chiamate al modello, anche quelle dei subagenti. Una strada gratuita non c'è: il piano Free non include Claude Code e l'API conteggia i token dalla prima richiesta. I crediti comprati direttamente nella Console di Anthropic sono prepagati, scadono un anno dopo l'acquisto e non sono rimborsabili (Anthropic help center, letto il October 3, 2026).
The least binding option is a prepaid key: on Kunavo, you can top up from $10, with no monthly fee, and the balance never expires. With $10, on Claude Sonnet 5 you can complete about 230 agent steps, or approximately 12 activities of 20 steps (not counting cache).
Claude Code via API: prices per token
Con una chiave non ci sono piani né finestre di utilizzo: Claude Code viene conteggiato come qualsiasi altro traffico API, a token. File e risultati degli strumenti contano come input; modifiche, spiegazioni e token di ragionamento (thinking) come output. Prezzi in dollari per milione di token (input / output); listino Anthropic da claude.com/pricing (letto il October 3, 2026), prezzi Kunavo dal catalogo (aggiornati al October 3, 2026):
| Model | Anthropic list price | Kunavo | Savings | One step | In Claude Code for… |
|---|---|---|---|---|---|
Claude Haiku 4.5claude-haiku-4-5 | $1,00 / $5,00 | $0,70 / $3,50 | 30% | $0,0217 | Background tasks, summaries, simple subagents |
Claude Sonnet 4.6claude-sonnet-4-6 | $3,00 / $15,00 | $2,10 / $10,50 | 30% | $0,0651 | Previous generation, for existing configurations |
Claude Sonnet 5claude-sonnet-5 | $2,00 / $10,00 | $1,40 / $7,00 | 30% | $0,0434 | Everyday work (starting model in the configuration below) |
Claude Opus 5claude-opus-5 | $5,00 / $25,00 | $3,50 / $17,50 | 30% | $0,1085 | Planning and edits across multiple files |
Claude Opus 5.5claude-opus-5-5 | $4,00 / $20,00 | $2,80 / $14,00 | 30% | $0,0868 | The latest Opus (opus alias below); reasoning cannot be disabled |
A step is an exchange with the model within an activity: system prompt, history, and fresh context from files as input; an edit or explanation as output—approximately 25.000 input tokens and 1.200 output tokens, without cache. All language versions of this guide use the same unit, starting with the English version, Claude Code pricing.
The cache determines the bill
At every step, Claude Code resends the entire conversation: much of the input is therefore repeated context, which prompt caching makes much cheaper. Here is how Kunavo counts the cache (per million tokens, 5-minute window; the multiplier applies to the model's input price):
| Model | Cache reading | Cache writing | Input without cache |
|---|---|---|---|
| Claude Haiku 4.5 | 0,1× = $0,07 | 1,25× = $0,875 | $0,70 |
| Claude Sonnet 4.6 | 0,1× = $0,21 | 1,25× = $2,625 | $2,10 |
| Claude Sonnet 5 | 0,1× = $0,14 | 1,25× = $1,75 | $1,40 |
| Claude Opus 5 | 0,1× = $0,35 | 1,25× = $4,375 | $3,50 |
| Claude Opus 5.5 | 0,05× = $0,14 | 1,25× = $3,50 | $2,80 |
Watch out for the long coffee. La cache non è eterna: con una chiave API resta valida cinque minuti, mentre l'abbonamento la tiene un'ora (Cost documentation, letta il October 3, 2026). Torni alla scrivania dopo un quarto d'ora e scrivi il prossimo messaggio? Quel contesto non viene più letto dalla cache ma riscritto, alla tariffa di scrittura. Su una conversazione da 200.000 token con Claude Sonnet 5 la differenza su Kunavo è tra circa $0,028 e circa $0,35, per un solo messaggio, e il conto si ripete a ogni pausa. Se al rientro cambi comunque argomento, un /clear prima di ricominciare evita di pagare un contesto che non ti serve più. I dettagli tecnici sono nella Cache documentation (in inglese).
Example session: how much it costs with caching
Qui sotto un calcolo che puoi rifare. Il punto di partenza è l'output di esempio del comando /usage nella Anthropic cost documentation (letta il October 3, 2026). Le ipotesi:
- The token counts come from the documentation: 1.200 input, 5.300 output, 940.000 read from cache, and 50.000 written to cache. For Anthropic, it is a session on Sonnet 4.6, which costs $0.55 at list price.
- We apply the same counts to the rates of four models; only the price changes. Tokenizer differences between models are ignored.
- Cache writing is calculated at the 5-minute window rate, the default duration with an API key.
- Reasoning tokens are paid for as output (same documentation): more reasoning increases the output portion.
- Background calls are excluded; according to the documentation, they usually cost less than $0.04 per session.
| Model | Anthropic list price | Kunavo | Share of cache reading in the Kunavo amount |
|---|---|---|---|
| Claude Sonnet 5 | $0,368 | $0,258 | approximately 51% |
| Claude Opus 5.5 | $0,549 | $0,384 | approximately 34% |
| Claude Opus 5 | $0,921 | $0,645 | approximately 51% |
| Claude Haiku 4.5 | $0,184 | $0,129 | approximately 51% |
On Claude Sonnet 5, this session costs approximately $0,258 on Kunavo and approximately $0,368 at Anthropic's list price; cache reading accounts for approximately 51% of the Kunavo amount. With $20, the Pro price, you can run approximately 77 sessioni sessions like this on Kunavo. Claude Opus 5.5 in the same example costs approximately $0,384, Claude Opus 5 approximately $0,645, and Claude Haiku 4.5 approximately $0,129.
This is arithmetic based on an example, not a measured bill or a spending cap: your sessions may be much longer or shorter. In Claude Code, /usage shows the four counts for the current session—enter them in this script:
# Costo di una sessione di Claude Code dai suoi conteggi di token (USD, prezzi Kunavo).
# In Claude Code, /usage mostra i quattro numeri della sessione in corso.
TARIFFE = { # per 1M di token: (input, output, lettura cache, scrittura cache)
"claude-sonnet-5": (1.40, 7.00, 0.14, 1.75),
"claude-opus-5-5": (2.80, 14.00, 0.14, 3.50),
"claude-opus-5": (3.50, 17.50, 0.35, 4.375),
"claude-haiku-4-5": (0.70, 3.50, 0.07, 0.875),
}
def costo_sessione(modello, token_input, token_output, letti_cache, scritti_cache):
i, o, r, w = TARIFFE[modello]
return (token_input * i + token_output * o + letti_cache * r + scritti_cache * w) / 1e6
# La sessione di esempio della documentazione sui costi di Anthropic:
print(round(costo_sessione("claude-sonnet-5", 1_200, 5_300, 940_000, 50_000), 3)) # 0.258The dollar amount shown by /usage is an estimate that Claude Code calculates locally using Anthropic’s list prices; with Kunavo, you can see the amount actually deducted in the dashboard’s usage history. You can calculate your scenarios in the Claude token cost calculator.
For a sense of scale: according to the cost documentation, in enterprise deployments the average cost is about $13 per developer per active working day and $150–250 per developer per month, and it remains below $30 per active day for nine out of ten users. These are Anthropic enterprise averages, not Kunavo prices.
Subscription or API: which should you choose?
The two options on the same scale: monthly fee ÷ cost of one step = number of steps per month above which the subscription pays off. At Anthropic prices, excluding taxes and without caching:
| Plan | Fee (excluding taxes) | Steps on Claude Sonnet 5 with Kunavo ($0,0434 per step) |
|---|---|---|
| Pro | $20 per month | 460 steps (approximately 23 activities of 20 steps) |
| Max 5x | $100 per month | 2.300 steps (approximately 115 activities of 20 steps) |
| Max 20x | $200 per month | 4.610 steps (approximately 230 activities of 20 steps) |
With a warm cache—when 20.000 of the 25.000 input tokens come from the cache and the rest are written—one step costs about $0,020, so $20 already gets you about 1.000. An intensive month (20 days at 100 steps) on Claude Sonnet 5 costs about $87 on Kunavo without caching and about $40 with a warm cache. Anthropic does not publish how many steps fit within its plan limits; the comparison only indicates above how many steps the fixed fee costs less, provided the limits allow it.
Translated into concrete profiles:
- Full-time developer, with Claude Code open all day: about 5 giorni of intensive work (approximately 460 steps) is enough to spend as much with the key as Pro costs for one month. If Pro’s limits are sufficient, choose Pro.
- Considering Max 5x: on Claude Sonnet 5, the fee equals about 2.300 steps, or roughly 23 giorni full working days. It makes sense only if Claude Code is your primary tool for much of the month.
- Freelancers, students, weekend projects: you usually stay below these figures, and the key costs less because an inactive month costs $0. The same applies to CI pipelines, automated scripts, and teams that want a spending cap for each person.
- Already have Pro and occasionally hit the limit: keep the subscription and use the key only during the hours when the plan is blocked, instead of buying usage credits at full API rates.
Is Max worth it?
The right question is not “Max or Pro?” but “How much work do I lose when Pro stops?” Upgrading to Max 5x costs $80 more per month. Suppose the limit blocks you for an entire afternoon, say 50 steps on Claude Sonnet 5: continuing with the Kunavo key costs about $2,17, without caching. The $80 difference therefore covers about 37 pomeriggi of those (approximately 1.840 steps). If this happens less often, stay on Pro and keep the key as a backup; if it happens almost every day, Max is the simpler choice.
Configuring Claude Code with Kunavo
Claude Code reads ANTHROPIC_BASE_URL, so it connects to Kunavo through Anthropic’s Messages API, without additional plugins. For a single terminal session, or permanently in ~/.zshrc:
export ANTHROPIC_BASE_URL=https://api.kunavo.com # solo l'indirizzo, senza /v1
export ANTHROPIC_AUTH_TOKEN=sk-kn-...
export ANTHROPIC_MODEL=claude-sonnet-5
export ANTHROPIC_DEFAULT_OPUS_MODEL=claude-opus-5-5
export ANTHROPIC_DEFAULT_SONNET_MODEL=claude-sonnet-5
export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-4-5ANTHROPIC_BASE_URL è solo l'indirizzo del server: il percorso /v1/messages lo aggiunge Claude Code. Usa ANTHROPIC_AUTH_TOKEN e non ANTHROPIC_API_KEY: la prima variabile finisce nell'intestazione Authorization: Bearer, la seconda in x-api-key (Documentation on connecting to a gateway, letta il October 3, 2026). Kunavo accetta la chiave in entrambe le intestazioni, ma ANTHROPIC_AUTH_TOKEN vale subito, mentre ANTHROPIC_API_KEY in modalità interattiva va approvata una volta, e una chiave rifiutata in precedenza Claude Code la ignora senza chiedere di nuovo. Conviene mettere le variabili nel blocco env delle impostazioni utente; se la stessa variabile è sia nella shell sia nel file di impostazioni, vince il file:
{
"env": {
"ANTHROPIC_BASE_URL": "https://api.kunavo.com",
"ANTHROPIC_AUTH_TOKEN": "sk-kn-...",
"ANTHROPIC_MODEL": "claude-sonnet-5",
"ANTHROPIC_DEFAULT_OPUS_MODEL": "claude-opus-5-5",
"ANTHROPIC_DEFAULT_SONNET_MODEL": "claude-sonnet-5",
"ANTHROPIC_DEFAULT_HAIKU_MODEL": "claude-haiku-4-5"
}
}Never write the key in the project’s .claude/settings.json file: that file ends up in the repository. In the VS Code extension, set the variables in claudeCode.environmentVariables, in VS Code’s user settings.
Why we pin the models. Secondo la model configuration documentation (letta il October 3, 2026), sull'API Anthropic Claude Code parte di default con Opus 5.5, l'alias opus punta a Opus 5.5 e l'alias sonnet a Sonnet 5.5. Sonnet 5.5 is not available on Kunavo: without ANTHROPIC_DEFAULT_SONNET_MODEL, the /model sonnet command, the opusplan execution phase, and subagents configured for sonnet would request a model that Kunavo does not serve and would receive a 404 error. ANTHROPIC_MODEL imposta come modello di partenza Claude Sonnet 5, più economico; ANTHROPIC_DEFAULT_SONNET_MODEL fa rispondere lo stesso modello all'alias sonnet, e ANTHROPIC_DEFAULT_HAIKU_MODEL decide anche il modello delle funzioni in background. L'alias opus lo fissiamo su Claude Opus 5.5 ($2,80 / $14,00), così resta su un modello servito da Kunavo anche quando Anthropic lo sposterà su un Opus più recente. Serve Claude Code v2.1.280 o successivo: con una versione precedente esegui claude update. Su questo modello il ragionamento non si può disattivare, e i token di ragionamento si pagano come output (Cost documentation).
Start claude and type /status: the Anthropic base URL line should show https://api.kunavo.com, and the Auth token line should show the ANTHROPIC_AUTH_TOKEN variable. If you instead see the claude.ai account login, the variable did not reach the session. The complete guide (in English): Claude Code with Kunavo. Create the key, which starts with sk-kn-, after signing up on the keys page.
Paying with Satispay: only for the API key
Kunavo è un gateway API compatibile con le interfacce di Anthropic e di OpenAI. La ricarica del saldo passa dal checkout di Stripe. I prezzi sono in dollari, ma se paghi dall'Italia Stripe ti mostra l'importo in euro prima che tu confermi. Secondo la Stripe documentation on Adaptive Pricing il cambio include una commissione di conversione del 2–4%, a carico di chi paga; se scegli di pagare in dollari non la paghi, ma possono applicarsi il cambio e le spese della tua banca (letto il October 3, 2026). Satispay però funziona solo con il checkout in euro, quindi una ricarica con Satispay include sempre quella commissione. Con il checkout in euro trovi:
- Satispay, offerto al checkout in euro. Scelto Satispay, inserisci il numero di cellulare collegato al tuo account oppure inquadri il QR code con l'app, poi tocchi Submit (Satispay support, letto il October 3, 2026). Stripe lo presenta solo in euro (Stripe documentation on Satispay).
- Cards (Visa, Mastercard) and Link.
- Apple Pay in Safari and on iPhone, Google Pay in Chrome and on Android, if you have configured Google Pay.
Mancano PayPal, l'addebito diretto SEPA e il bonifico; Klarna è offerto solo a chi paga dagli Stati Uniti. Pay by Bank dall'Italia non c'è: secondo la Stripe documentation (letta il October 3, 2026) è disponibile solo per chi paga da Regno Unito, Irlanda e Finlandia, mentre in Francia e Germania è in anteprima privata. Kunavo non emette fatture; lo storico delle ricariche (data, importo, stato) è nella pagina Billing della dashboard.
For transparency: Satispay was added to Kunavo’s checkout on October 3, 2026. Stripe offers it for euro payments, but so far no top-up has been made through Satispay. If it does not appear at your checkout, cards, Apple Pay, Google Pay, and Link remain alternatives.
- Minimum top-up $10; higher amounts receive a bonus: $100 → $110 di saldo, $1.000 → $1.200 di saldo, $5.000 → $6.250 di saldo.
- The balance is prepaid, has no monthly fee, and does not expire; failed requests are not charged.
- La ricarica automatica funziona solo con una carta salvata o con Link. Nel checkout di Stripe Satispay è un metodo a uso singolo (Stripe documentation), quindi ogni ricarica con Satispay la fai a mano.
- These methods top up your Kunavo API balance. They do not pay for Claude Pro or Max: the Anthropic website accepts cards only.
The step-by-step procedure is in the guide paying for Claude with Satispay. To get started: create your Kunavo account, top up the balance, and create a key. Prices for all models are in the Kunavo price list.
What you lose by switching to the key
Poco, ma qualcosa sì. Tutto ciò che passa dall'API Messages continua a funzionare uguale: l'agente che legge e modifica i file, i comandi nel terminale, i subagenti, i server MCP. Non funzionano invece le due funzioni legate al tuo account claude.ai, cioè Remote Control e la dettatura vocale: con una credenziale di gateway attiva spariscono, e Remote Control resta spento anche quando ANTHROPIC_BASE_URL punta a un server diverso da quello di Anthropic (Documentation on connecting to a gateway, letta il October 3, 2026). Se le usi spesso, è un motivo concreto per restare sull'abbonamento.
Va detto anche che Kunavo è un servizio indipendente e serve Claude su capacità condivisa: Anthropic dichiara di non approvare, mantenere o verificare i gateway di terze parti (LLM gateway documentation). In sintesi: Pro per chi passa le giornate dentro Claude Code e non tocca il limite, la chiave per tutto il resto.
Frequently asked questions
How much does Claude Code cost?
Claude Code itself is free: you pay for the model, either through a subscription or based on usage. With a subscription, Claude Code is included in Claude Pro ($20 per month, or $17 per month with the annual subscription) and Claude Max ($100 or $200 per month), as well as Team and Enterprise plans. These are Anthropic's dollar prices, excluding taxes, according to claude.com/it/pricing (accessed on October 3, 2026). With an API key, you pay for tokens with no subscription fee: on Kunavo, Claude Sonnet 5 costs $1,40 / $7,00 per million input and output tokens, so one agent step (approximately 25.000 input tokens and 1.200 output tokens) costs about $0,0434.
Is Claude Code free?
Only the tool. The CLI, editor extensions, and configuration cost nothing; model calls do. According to Anthropic's documentation, the free claude.ai plan does not include Claude Code, and the API counts tokens from the first request. Without a subscription, you can use a prepaid key: on Kunavo, top-ups start at $10, the balance does not expire, and a month without programming costs $0.
How much does Claude Pro cost in euros, including VAT?
Anthropic lists Pro's price in dollars: $20 per month in the United States, excluding taxes. According to its help center, where supported, payment is made in the local currency; monthly prices vary by region, and in some regions the tax is already included in the displayed price, while in others it is added at payment. The standard VAT rate in Italy is 22% (European Commission, Your Europe), and according to Anthropic, the billing address determines the tax jurisdiction. The euro figures in Italian guides do not match; the amount shown at payment applies. A subscription purchased in the app is charged by Apple or Google, and the price may differ.
Can I pay for Claude Code with Satispay or PayPal?
For a Claude Pro or Max subscription on claude.ai, no: according to Anthropic's payment FAQs, subscriptions purchased on the web can be paid for only with a credit or debit card, and Anthropic does not accept third-party processors such as PayPal. With Satispay, you pay for the API route: Kunavo balance top-ups go through Stripe's checkout, which offers Satispay, cards (Visa, Mastercard), Link, Apple Pay, and Google Pay when the amount is in euros. Stripe's conversion includes a 2–4% conversion fee. PayPal and SEPA direct debit are unavailable, Klarna is reserved for customers paying from the United States, and Kunavo does not issue invoices. The minimum top-up is $10.
Subscription or API key: which is better?
Divide the monthly subscription fee by the cost of one agent step. On Kunavo, a step with Claude Sonnet 5 costs about $0,0434, so Pro's $20 equals about 460 steps per month (approximately 23 activities of 20 steps), and Max 5x's $100 equals about 2.300 steps. If you exceed that amount every month and Pro's limits do not hold you back, choose Pro. If you work in phases, the key costs less: in months when you do not use it, you pay $0.
Is the API included in Claude Pro?
No. The Claude subscription and Claude API are separate products with separate payments. If you set ANTHROPIC_AUTH_TOKEN or ANTHROPIC_API_KEY in Claude Code, Claude Code uses the key instead of the subscription: the plan limits no longer apply, and the key holder pays for the tokens. For a Kunavo key, also set ANTHROPIC_BASE_URL=https://api.kunavo.com (without /v1); ANTHROPIC_AUTH_TOKEN works immediately, while ANTHROPIC_API_KEY in interactive mode must be approved once. The claude.ai login remains saved: remove the variable and Claude Code returns to the subscription. The /status command shows which of the two is active.
How much does a Claude Code session via the API cost?
It depends on the tokens. The example session in Anthropic's cost documentation (1.200 input tokens, 5.300 output tokens, 940.000 read from cache, and 50.000 written to cache, recorded on Sonnet 4.6) at rates of Claude Sonnet 5 costs about $0,258 on Kunavo and about $0,368 at Anthropic's list price; cache reading accounts for approximately 51% of the Kunavo amount. This is a recalculation of the example, not a measured bill: tokenizer differences are ignored. Your session counts are shown by the /usage command in Claude Code.
What do I do when Claude Code's limit runs out?
With a subscription, work stops until reset: the session limit resets every five hours, and the weekly limit resets once a week. According to Claude Code's cost documentation, both apply to all models together, so switching models with /model does not help; it helps only if you have exhausted a model-specific limit. Three options remain: wait, move to a higher plan, or continue on a paid basis by activating usage credits (prepaid, at standard API rates) or setting your own API key, which Claude Code uses while the variable is set.