Claude APIEvery Claude model, 30% under Anthropic's list price
Kunavo is an independent API gateway that resells access to Claude. It is not Anthropic. One key calls the Claude models below in Anthropic's Messages format or through the OpenAI SDK, and the cost of each call is deducted per token from a prepaid balance.
import os
from anthropic import Anthropic
# Pass the origin: the SDK appends /v1/messages itself.
client = Anthropic(
api_key=os.environ["KUNAVO_API_KEY"],
base_url="https://api.kunavo.com",
)
message = client.messages.create(
model="claude-sonnet-5",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello, Claude"}],
)
for block in message.content:
if block.type == "text":
print(block.text)For the Anthropic SDK the base URL is https://api.kunavo.com, with no /v1: the SDK adds the path itself. The key comes from /app/keys, and the model IDs are the ones in the table below.
What each Claude model costs, next to Anthropic's list price
US dollars per 1M tokens. In each cell the first figure is Kunavo's price and the figure under it is Anthropic's list price for the same kind of token.
| Model | Input tokens | Output tokens | Cache read | Cache write | Below list price |
|---|---|---|---|---|---|
| Claude Fable 5.1 claude-fable-5-1 | $7.00 Anthropic $10.00 | $35.00 Anthropic $50.00 | $0.175 Anthropic $0.25 | $8.75 Anthropic $12.50 | −30% |
| Claude Fable 5 claude-fable-5 | $7.00 Anthropic $10.00 | $35.00 Anthropic $50.00 | $0.70 Anthropic $1.00 | $8.75 Anthropic $12.50 | −30% |
| Claude Opus 5.5 claude-opus-5-5 | $2.80 Anthropic $4.00 | $14.00 Anthropic $20.00 | $0.14 Anthropic $0.20 | $3.50 Anthropic $5.00 | −30% |
| Claude Opus 5 claude-opus-5 | $3.50 Anthropic $5.00 | $17.50 Anthropic $25.00 | $0.35 Anthropic $0.50 | $4.375 Anthropic $6.25 | −30% |
| Claude Opus 4.8 claude-opus-4-8 | $3.50 Anthropic $5.00 | $17.50 Anthropic $25.00 | $0.35 Anthropic $0.50 | $4.375 Anthropic $6.25 | −30% |
| Claude Opus 4.7 claude-opus-4-7 | $3.50 Anthropic $5.00 | $17.50 Anthropic $25.00 | $0.35 Anthropic $0.50 | $4.375 Anthropic $6.25 | −30% |
| Claude Opus 4.6 claude-opus-4-6 | $3.50 Anthropic $5.00 | $17.50 Anthropic $25.00 | $0.35 Anthropic $0.50 | $4.375 Anthropic $6.25 | −30% |
| Claude Sonnet 5 claude-sonnet-5 | $1.40 Anthropic $2.00 | $7.00 Anthropic $10.00 | $0.14 Anthropic $0.20 | $1.75 Anthropic $2.50 | −30% |
| Claude Sonnet 4.6 claude-sonnet-4-6 | $2.10 Anthropic $3.00 | $10.50 Anthropic $15.00 | $0.21 Anthropic $0.30 | $2.625 Anthropic $3.75 | −30% |
| Claude Haiku 4.5 claude-haiku-4-5 | $0.70 Anthropic $1.00 | $3.50 Anthropic $5.00 | $0.07 Anthropic $0.10 | $0.875 Anthropic $1.25 | −30% |
The cache-write price is for a cache write kept for 5 minutes.
Prices on this page are read live from the Kunavo model catalog, so they match what the API bills.
Worked costs for real workloads are in the Claude API pricing guide, and the Claude token cost calculator prices a workload of your own.
Already on the OpenAI SDK? Keep it.
The same Claude models are served on /v1/chat/completions. Point the OpenAI SDK at https://api.kunavo.com/v1, set model to a Claude model ID, and the call is billed at the same prices as a call in the Messages format.
import os
from openai import OpenAI
# The base URL ends in /v1: the SDK appends /chat/completions.
client = OpenAI(
api_key=os.environ["KUNAVO_API_KEY"],
base_url="https://api.kunavo.com/v1",
)
response = client.chat.completions.create(
model="claude-sonnet-5",
messages=[{"role": "user", "content": "Hello, Claude"}],
)
print(response.choices[0].message.content)The Claude models you can call today
The ID under each name is what goes in model. Each name opens that model's page, with its context window, its output limit and a request you can copy.
- Claude Fable 5.1
claude-fable-5-1Anthropic's most capable model — frontier reasoning and long-horizon agentic work.
1M context window · 128K max output - Claude Fable 5
claude-fable-5Frontier reasoning and long-horizon agents — the model Fable 5.1 succeeds.
1M context window · 128K max output - Claude Opus 5.5
claude-opus-5-5Anthropic's newest Opus — long-running agentic coding and knowledge work, for less than Opus 5.
1M context window · 128K max output - Claude Opus 5
claude-opus-5Near-flagship Opus reasoning at half the price of Fable 5 — vision and agentic coding.
1M context window · 128K max output - Claude Opus 4.8
claude-opus-4-8Anthropic Opus 4.8 — stronger agentic coding and honesty.
1M context window · 128K max output - Claude Opus 4.7
claude-opus-4-7Anthropic Opus 4.7 — previous-generation Opus; reasoning and vision.
1M context window · 128K max output - Claude Opus 4.6
claude-opus-4-6Anthropic Opus 4.6 — deep reasoning, exceptional agentic ability.
1M context window · 128K max output - Claude Sonnet 5
claude-sonnet-5Near-Opus coding and agentic quality at Sonnet cost.
1M context window · 128K max output - Claude Sonnet 4.6
claude-sonnet-4-6Balanced speed/quality — the everyday production workhorse, elite coding.
1M context window · 128K max output - Claude Haiku 4.5
claude-haiku-4-5Anthropic Haiku 4.5 — fast and cost-efficient.
200K context window · 64K max output
Single upstreamServed through one supplier today. If that supplier has an outage, calls fail until it recovers — failed calls are never billed. We're adding a second supplier to every text model.
Claude Code, and the clients you already use
Two environment variables — ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN — move Claude Code onto pay-as-you-go without changing the install.
# One terminal session
export ANTHROPIC_BASE_URL=https://api.kunavo.com # origin, no /v1
export ANTHROPIC_AUTH_TOKEN=sk-kn-...
# Pin models Kunavo serves. Claude Code's default and its opus/sonnet aliases
# follow Anthropic's newest models; the sonnet alias asks for Sonnet 5.5,
# which Kunavo does not serve — unpinned, /model sonnet 404s.
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-4-5
claudeCoding agents, editors and CLIs
- OpenClaw
- Hermes Agent
- Claude Code
- Kilo Code
- Roo Code
- Cline
- Codex CLI
- opencode
- Claude Code Router
- Zed
- Continue
- Aider
- CC Switch
- Claude Agent SDK
- Amp
- GitHub Copilot CLI
- Crush
- DeepSeek Harness
- Dyad
- Factory Droid
- goose
- Jan Agent
- Junie CLI
- Kimi Code
- mini-SWE-agent
- Mistral Vibe CLI
- Msty Go
- Nanocoder
- Oh My Pi
- Open Interpreter
- OpenHands
- Pi
- Qoder
- Qwen Code
- Theia IDE
- twinny
Each name opens the setup page for that client. A client that is not listed works if it has a field for a base URL: all integrations.
Prepaid from $10, with the payment methods of your country
There is no subscription and no monthly fee. You top up from $10, and the cost of each call is deducted from that balance. The balance does not expire, and a call that fails is not billed.
- Everywhere Stripe works: cards (Visa, Mastercard, Amex, JCB, UnionPay). Outside India also Link (Stripe's wallet), and Apple Pay or Google Pay on a device that has it set up.
- United States: Cash App Pay, Klarna, Afterpay, ACH bank debit and crypto (stablecoins). Affirm is offered on top-ups from $100, and Zip on top-ups from $100 to $1,500.
- Mainland China: Alipay and WeChat Pay.
- South Korea: KakaoPay, Naver Pay, PAYCO, Samsung Pay and Korean domestic cards.
- India and Brazil: UPI in India, Pix in Brazil.
- Europe: iDEAL in the Netherlands, Bancontact in Belgium, EPS in Austria, MB WAY in Portugal, Satispay in Italy and BLIK in Poland.
- United Kingdom, Ireland and Finland: Pay by Bank; in the United Kingdom also Bacs Direct Debit.
Top-up amounts are in US dollars. Where the checkout shows your own currency, Stripe's exchange rate includes a 2–4% conversion fee, which paying in dollars avoids. The conditions of each method are in the billing documentation, and the guide to paying without a credit card covers the methods country by country.
Kunavo vs calling Anthropic directly
What changes when Claude is called through a gateway instead of Anthropic's own API, including where Anthropic's own API is the better choice.
| Kunavo | Anthropic direct | |
|---|---|---|
| Price | 30% under Anthropic's list price on every Claude model | Anthropic's list price |
| Account | One Kunavo account, key in 2 minutes, $10 minimum top-up | A Anthropic account plus its own billing setup |
| SDK | Keep the OpenAI SDK — change base_url, set model to "claude-sonnet-5" | Anthropic's SDK, or their OpenAI-compatible layer where offered |
| Same key also reaches | Claude, GPT, image, video and audio models | Anthropic's own catalog only |
| Billing | One Stripe balance, pay-as-you-go, never expires; failed requests unbilled | A separate invoice per provider |
| Rate limits | Shared gateway capacity, no contractual per-account limit | Anthropic's own tier limits, raised by usage history or contract |
Questions about the Claude API on Kunavo
Is this Anthropic's official Claude API?
No. Kunavo is an independent gateway that resells access to Claude models. Kunavo is not Anthropic, and it does not train or host the models. Anthropic's own API is at api.anthropic.com, with keys issued at platform.claude.com. A Kunavo key starts with sk-kn- and works only against api.kunavo.com.
How much does the Claude API cost on Kunavo?
On Kunavo, Claude is billed per token, at prices in US dollars per 1M tokens: Claude Sonnet 5 costs $1.40 per 1M input tokens and $7.00 per 1M output tokens. Anthropic's list price for the same model is $2.00 for input and $10.00 for output. Every Claude model on Kunavo is priced 30% under Anthropic's list price, and the table on this page gives the input, output, cache-read and cache-write price of each one.
Do the Anthropic SDK and Claude Code work with a Kunavo key?
Yes. Kunavo serves /v1/messages in Anthropic's Messages format, so an Anthropic SDK works once its base URL is https://api.kunavo.com and its API key is a Kunavo key. Claude Code takes the same two settings as environment variables, ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN, and you set its model variables to IDs that Kunavo serves.
Can I call Claude with the OpenAI SDK?
Yes. Set the OpenAI SDK's base URL to https://api.kunavo.com/v1 and the model to a Claude model ID such as claude-sonnet-5. The request goes to /v1/chat/completions and is billed at the same prices as a request in the Messages format.
Which Claude models can I call through Kunavo?
Today: Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Opus 4.6, Claude Sonnet 5, Claude Sonnet 4.6, and Claude Haiku 4.5. GET /v1/models returns the models a key can call at that moment, and a model that is not being served is left out of that list.
Does prompt caching work with Claude on Kunavo?
Yes. In the Messages format Kunavo passes the request on as it was sent, so a prefix marked with cache_control is cached. On the OpenAI-compatible endpoint Kunavo places the cache breakpoints itself on a long prompt. A cache read is billed at 10% of the model's input price (2.5% on Claude Fable 5.1, 5% on Claude Opus 5.5), and a cache write kept for 5 minutes at 1.25 times the input price.
How do I pay for the Claude API on Kunavo? Is there a subscription?
There is no subscription. You top up from $10, the amount becomes a prepaid balance, and the cost of each call is deducted from it per token. The balance does not expire, and a call that fails is not billed. Cards work wherever Stripe does. A checkout in a local currency also offers that country's own methods, among them Pix in Brazil, UPI in India, Alipay and WeChat Pay in mainland China, and KakaoPay in South Korea.
When is Anthropic's own API the better choice?
Go direct to Anthropic when you need a contractual rate limit, an enterprise SLA, or a provider-only feature on day one. Use Kunavo when you want one key, one balance and a lower per-token rate across every provider.
Three minutes to your first call.
One OpenAI-compatible API for Claude, GPT, Nano Banana, Veo and Suno — $10 minimum top-up, pay only for what you call.