Recharge credits for an OpenAI-compatible relay

NexRelay is prepaid. Credits are consumed by request-level input, output, and cached read token meters. When available balance reaches zero, requests stop with 402 until you recharge or add credit.

Prepaid only

No monthly overage and no surprise billing. Balance must exist before upstream routing.

Request-level ledger

Every completed deduction is tied to a request id, model, token meter, unit price, and status.

Budget controls

Per-key daily budget, monthly budget, model allowlist, IP allowlist, and max output caps are enforced before routing.

Starter

For prototypes, small tools, and light API usage.

$9recharge

9 USD of prepaid wallet credit

Input price: $1.00 / 1M
Output price: $1.00 / 1M
Cache price: $0.10 / 1M
Included input: 3M tokens
Included output: 600K tokens
Included cache read: 6M tokens
Max context: 32K
Max output: 1K
Rate limit: 20 RPM

Default route: gpt-5.2 -> gpt-5.3 -> gpt-5.4 -> gpt-5.5

IncludedAll GPT versions available with cost controls
Included32K max context and 1K max output tokens

PayPal setup is pending.

Prepaid only. When available balance reaches zero, API requests are rejected with 402 until you recharge or add credit.

Preferred

Pro

For developers and teams running production workflows.

$29recharge

29 USD of prepaid wallet credit

Input price: $1.00 / 1M
Output price: $1.00 / 1M
Cache price: $0.10 / 1M
Included input: 18M tokens
Included output: 4M tokens
Included cache read: 40M tokens
Max context: 128K
Max output: 4K
Rate limit: 120 RPM

Default route: gpt-5.3 -> gpt-5.2 -> gpt-5.4 -> gpt-5.5

IncludedProduction routing across GPT-5.2 to GPT-5.5
Included128K max context and 4K max output tokens

PayPal setup is pending.

Prepaid only. When available balance reaches zero, API requests are rejected with 402 until you recharge or add credit.

Business

For higher concurrency, longer context, and priority routing.

$99recharge

99 USD of prepaid wallet credit

Input price: $1.00 / 1M
Output price: $1.00 / 1M
Cache price: $0.10 / 1M
Included input: 85M tokens
Included output: 18M tokens
Included cache read: 200M tokens
Max context: 256K
Max output: 8K
Rate limit: 600 RPM

Default route: gpt-5.4 -> gpt-5.5 -> gpt-5.3 -> gpt-5.2

IncludedPriority routing for high-value workloads
Included256K max context and 8K max output tokens

PayPal setup is pending.

Prepaid only. When available balance reaches zero, API requests are rejected with 402 until you recharge or add credit.

Token meters

Meter	Unit price	Billing note
Input tokens	$1.00 / 1M metered tokens	Billed from reported request usage.
Output tokens	$1.00 / 1M metered tokens	Billed from reported request usage.
Cached read tokens	$0.10 / 1M metered tokens	Cache hits are billed at a lower weighted rate.

Deduction example

A request with 100,000 reported input tokens, 20,000 output tokens, and 50,000 cached read tokens bills 50,000 non-cached input tokens plus cache at the lower rate. Total deduction: 75,000 credits, equal to $0.0750.

Formula: non-cached input tokens + output tokens + cached read tokens x 0.1, converted into prepaid wallet credits.

Billing rules

Each API request has one request id and can be billed at most once.
Balance is checked before the upstream request. Insufficient balance returns 402 and no upstream request is sent.
Successful requests are billed from actual non-cached input, output, and cached read token usage when upstream usage is available.
Upstream failures before a usable response are not billed. Partial successful streamed responses can be billed by reported usage.
Retries and abnormal traffic are logged; repeated failures or high retry rates may be rate-limited or reviewed.

FAQ

What happens when credits run out?

The gateway returns 402 before sending the request upstream. Recharge or add a credit pack to continue using the API.

Are failed requests billed?

Balance and budget rejections are not billed. Upstream failures before a usable response are not billed. Partial successful streamed responses can be billed when usage is reported.

Can I prevent a key from overspending?

Yes. Set daily budget, monthly budget, model allowlist, IP allowlist, and max output tokens on each API key.

Can I audit charges?

Yes. Usage Logs shows request id, key, model, token meters, unit prices, billed amount, and status, with CSV export.

OpenAI-compatible API surface

Unified wallet and usage reconciliation

PayPal checkout with prepaid credits