Cheap Claude API Access With Alipay and No US Card (2026)

If you are building outside the United States, the official Claude API can be hard to reach: it wants a provider account and, for most people, a US or international credit card that their bank happily lets through for a US SaaS charge. Plenty of developers in Southeast Asia, South Asia, Latin America and the Middle East have the budget but not the payment rails.

Updated 2026-10-10. Claude models on cloudgpu.app are currently available through Claude Code only. The setup below uses Claude Code.

CloudGPU runs a gateway that gives you Claude Opus and Sonnet in Claude Code on one prepaid balance, at half the official Claude API price (50% off), payable with the methods your card issuer actually approves — Alipay, Google Pay, Apple Pay, or USDT. No separate account, no US card.

Claude API prices, 8 October 2026

Per 1M tokens, in USD. “Ours” is what you are billed; “official list” is the published API price for the same model. Every row is exactly 50% of the official rate — half price.

Model	Input (ours / official list)	Output (ours / official list)
claude-opus-5-5	$2.00 / $4.00	$10.00 / $20.00
claude-opus-5	$2.50 / $5.00	$12.50 / $25.00
claude-sonnet-5-5	$1.00 / $2.00	$5.00 / $10.00
claude-sonnet-4-6	$1.50 / $3.00	$7.50 / $15.00
claude-fable-5-1	$5.00 / $10.00	$25.00 / $50.00

claude-opus-4-8, claude-opus-4-7 and claude-opus-4-6 match claude-opus-5’s rate; claude-sonnet-5 matches claude-sonnet-5-5; claude-fable-5 matches claude-fable-5-1. The authoritative, always-current table is generated from the billing feed on the /api page — if a number here ever disagrees with that page, the page is right.

Call it in 90 seconds

1. Sign up with an email or Google account at cloudgpu.app — no phone number.
2. Top up on the billing page: Alipay, Google Pay, Apple Pay, Visa or Mastercard from $5, or USDT (TRC-20). A first top-up of $10 or more gets a $5 bonus (top up $10, start with $15).
3. Create an API key in the API console. It starts with cgw-sk-.
4. Point Claude Code at the gateway and start it:

export ANTHROPIC_BASE_URL=https://cloudgpu.app
export ANTHROPIC_AUTH_TOKEN=cgw-sk-...
export ANTHROPIC_MODEL=claude-sonnet-5-5          # or claude-opus-5-5
export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-sonnet-5-5   # Haiku is not offered
claude

The base URL is the bare host, with no /v1. The same settings work in the Claude Code extensions for VS Code and JetBrains, and you can keep them in the env block of ~/.claude/settings.json. The full walkthrough is in Claude Code with a cheaper API endpoint.

Claude models are currently available through Claude Code only. Calling them from Cursor, Cline, Continue, the OpenAI SDK, the OpenAI-compatible /v1/chat/completions endpoint or curl is not supported right now and will fail. The same key and balance still work for every other model on the gateway (DeepSeek, GLM, Kimi, MiniMax and more) through the OpenAI-compatible endpoint https://cloudgpu.app/v1, in any of those tools.

Which Claude model to use

• claude-opus-5-5 — the strongest model here, for hard reasoning, large refactors and agentic coding. At $2.00 / $10.00 it is the cheapest way we know to run Opus-class work.
• claude-sonnet-5-5 — the everyday workhorse: fast, cheap ($1.00 / $5.00), strong enough for most coding, drafting and RAG. Start here and only move up if a task genuinely needs Opus.
• claude-sonnet-4-6 — a slightly older Sonnet kept for teams that pinned to it.
• claude-fable-5-1 / claude-fable-5 — the long-form writing line; higher priced because the list price is higher.

If you are unsure, send Sonnet the request first and escalate to Opus only when the answer is not good enough. The 2x price gap between them is the single biggest lever on your bill.

Where your requests go

The gateway runs in Hong Kong. There is no automatic failover right now, so an outage is an outage here — the same honest caveat we give for every model on the gateway. Prompt and completion bodies are not stored; we keep token counts, model, latency and error codes for billing only.

Two things to keep in mind. First, this is a gateway: you are buying API access billed from a prepaid balance, not a first-party provider account, so account-level features (a vendor console, workspaces, the vendor’s own rate-limit tiers) are not part of it. Second, the prices above are as of the date on this page and can change. The cheap, sane way to adopt any new provider is the same: make a small first top-up, run your real workload against it for a day, check latency and output quality, and only then load a larger balance.

Balance is prepaid, does not expire, and is refunded on request. Start at cloudgpu.app/api.