Documentation
Everything you need to plug CheapAI into your tooling — setup guides for popular agents, API reference, FAQ.
How is CheapAI different from Anthropic / OpenAI directly?
Same protocol, same response shape, same models — just cheaper and no VPN required. We don't re-label models under aliases and don't silently downgrade your traffic to «budget mode»: you pay for the exact tokens returned by the upstream provider.
How do you get these prices?
We operate as a reseller: we buy upstream access at tier-discounted rates and add a thin margin on top. Pricing is public and identical for everyone — no hidden surcharges and no «optimized» subscriptions.
Why does Claude Code's /cost show a huge number?
The built-in /cost uses official Anthropic pricing ($15/1M for Opus) — it doesn't know our rate. Look at the Logs page for the actual debits.
Why does each request «eat» 50-70k tokens?
That's how Claude Code and similar agents work: every turn they ship the full conversation + active files. Use /compact to shrink the history and start a new session for a new task.
How does caching work?
Anthropic supports prompt caching: repeated chunks (system prompts, documentation) are cached automatically. The first write is slightly more expensive (×1.25 of input), subsequent reads are much cheaper (×0.1 of input). On long sessions this saves 70-90% versus uncached runs.
Are you inflating token counts?
No. The API response includes a usage field with input_tokens / output_tokens — it comes straight from Anthropic / OpenAI / Google untouched. Easy to compare against an upstream-direct call.
What payment methods are supported?
Card (Visa / Mastercard) + PayPal via Lava (USD), SBP / card / SberPay (RUB) and crypto (USDT TRC20 / Bitcoin). Minimum top-up is $1 (USD card: $5; some coins, such as BTC and ETH, have a higher network minimum). Details on /topup.
How do I activate a promo code?
Open Top up → Promo code tab, paste the code and click «Activate» — balance updates instantly. Full step-by-step in the Promo code tab of these docs.
Where do I see the actual model list?
On the Models page — prices and context windows are listed per model. The list updates automatically when we add new providers or upstream changes rates.
Can I have a separate key per project?
Yes — create a key per project in Keys. That makes spend tracking easier and limits blast radius if one leaks.
How do I create an API key?
Sign up, open Keys, click Create, name it — the full string is shown once, copy it to a safe place immediately.