Pay 1 USD, get 5 USD credit

The Claude API,
5× the credit.

A fast, reliable gateway to Claude Opus 4.8, Sonnet 4.6, Haiku 4.5 and Fable 5. A drop-in replacement for the Anthropic SDK — just change the base URL. Pay-as-you-go, USDT accepted, no subscription.

bash
$ curl -X POST /v1/messages
# model: claude-opus-4-8
200 OK { "content": "Hello!" }
$

The full Claude model lineup

Call any Claude model through one API key — from the flagship Opus 4.8 to the ultra-fast Haiku 4.5, all with a 1M-token context window.

Flagship

Claude Opus 4.8

claude-opus-4-8

Anthropic’s most capable model for complex reasoning, agentic coding, and long-horizon tasks.

Reasoning Agentic coding 1M context
New

Claude Fable 5

claude-fable-5

A creative-writing specialist tuned for narrative, tone, and long-form prose.

Creative Writing 1M context

Claude Sonnet 4.6

claude-sonnet-4-6

The balanced workhorse — fast, smart, and cost-effective for everyday production traffic.

Balanced Everyday 1M context

Claude Haiku 4.5

claude-haiku-4-5

The fastest, lowest-cost Claude — ideal for high-volume, latency-sensitive workloads.

Fastest Low cost 1M context

Legacy Claude 3.x models remain available for backward compatibility.

A drop-in replacement for the Anthropic SDK

Keep your existing code. Point the base URL at XCMF, use your API key, and every Claude model works — including streaming, tools, and Claude Code, Cursor or any compatible client.

curl https://www.xcmfapi.com/v1/messages \
  -H "x-api-key: $XCMF_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello, Claude"}]
  }'

Pay 1 USD, get 5 USD of Claude credit

Usage-based pricing against official Anthropic rates — with a 5× credit ratio, no subscription, and balance that never expires.

You pay $1
You get $5

5× credit, billed at official rates

Top up 20 USD and get 100 USD of platform balance. Every request is metered per token against Anthropic's published Opus, Sonnet, and Haiku prices — so your effective cost is a fraction of buying direct.

Metered at official Anthropic per-token rates

USDT & crypto accepted

Top up with USDT — no credit card, no Anthropic account required. Built for overseas developers.

No subscription, no expiry

Pure pay-as-you-go. No monthly fee, and any balance you buy never expires.

Live in 5 minutes

Sign up, grab a key, change one base URL. Streaming and tools work out of the box.

Frequently asked questions

Everything developers ask about Claude API pricing, models, and getting started.

How much does the Claude API cost here?

You top up your balance and pay per token, priced against the official Anthropic rates. With our 5× credit ratio, every 1 USD you pay becomes 5 USD of platform balance — so effective per-token cost is a fraction of buying direct.

Is the Claude API free?

Anthropic has no free API tier. This gateway is pay-as-you-go with no subscription and no monthly fee — you only pay for the tokens you use, and unused balance never expires.

Which Claude model is best for coding?

Claude Opus 4.8 is the strongest for complex, agentic coding and long-horizon reasoning. Sonnet 4.6 is the balanced pick for most day-to-day coding at lower cost, and Haiku 4.5 is best for fast, high-volume tasks. All support up to a 1M-token context window.

What base URL should I use?

It is a drop-in replacement for the Anthropic SDK: keep your existing code and simply point the base URL to our endpoint, then use your XCMF API key. It also works with Claude Code, Cursor, and any OpenAI- or Anthropic-compatible client that lets you set a custom base URL.

Can I pay with USDT or crypto?

Yes. We support USDT (crypto) payments for overseas users. Add balance with USDT and start calling the Claude API within minutes — no credit card or Anthropic account required.

Do you support streaming responses?

Yes. Server-sent event (SSE) streaming is fully supported for the Messages API, so coding agents, IDEs, and production apps get token-by-token responses just like the official API.