Pricing
Up to 38x cheaper, in plain numbers
Every price below is blended USD per one million tokens. No tiers to decode, no per-seat fees, no monthly minimum.
| Model | ClaudeCheap price | Official price | Savings | Best for |
|---|---|---|---|---|
Opus claude-opus-4 | $0.40 / 1M | $15 / 1M | 38x cheaper | Hardest reasoning, long agentic runs, deep refactors. |
Sonnet claude-sonnet-4 | $0.30 / 1M | $3 / 1M | 10x cheaper | Everyday coding and chat. The default workhorse. |
Haiku claude-haiku-4 | $0.15 / 1M | $0.80 / 1M | 5x cheaper | High-volume, low-latency tasks and cheap classification. |
Hardest reasoning, long agentic runs, deep refactors.
Everyday coding and chat. The default workhorse.
High-volume, low-latency tasks and cheap classification.
Official prices are published Anthropic blended rates shown for comparison only. ClaudeCheap is not affiliated with Anthropic.
Billing
Pre-paid credits, metered per token
You are only ever billed for the tokens you actually send and receive.
Top up a balance
Add credit to your account. There is no subscription and no minimum spend.
Send requests
Route traffic through the proxy from any compatible tool or SDK.
Pay per token
We meter actual input and output tokens and deduct from your balance at the rate card price.
Billing FAQ
How payment works
How can it be this much cheaper?
We buy capacity in bulk and pass most of the savings on. You pay a published per-token rate that sits far below list price, with a thin margin that keeps the service running.
How does billing work?
Pre-paid credits. Top up a balance, then spend it down based on actual token usage at the rate card price. No seats and no monthly minimums.
What happens if my balance runs out?
Requests are rejected with a clear error until you top up again. Nothing is charged to a card automatically unless you opt into auto-reload.
Ready to wire it up?
Pick your tool and paste in one config block.