---
title: How to Buy a GPT API Key
description: "Buy a GPT API key without an OpenAI account: prepaid balance, card or crypto checkout, and GPT-5.6, GPT-5.5 and GPT Image 2 at 50% off official rates."
url: https://apitoken.sale/docs/learn/how-to-buy-gpt-api-key
language: en
---

# How to buy a GPT API key

To buy a GPT API key without an OpenAI Platform account, create an apiToken.sale account, top up a prepaid balance by card or crypto, and generate one sk-pool key. That key authenticates against an OpenAI-compatible endpoint with Authorization: Bearer and serves GPT-5.6, GPT-5.5 and GPT Image 2 at 50% off official token spend. This guide walks the purchase, the billing math and the exact client configuration.

## The purchase: account, balance, key

1. Create an apiToken.sale account with Google, GitHub, or email and password. Google and GitHub signups start with $5 of platform bonus credit, valid on supported Claude, GPT, Gemini and Kimi models; email/password accounts do not receive the bonus.
2. Top up any whole-dollar amount by bank card or cryptocurrency. There is no fixed bundle and no monthly commitment — the balance is prepaid and is spent only when requests run.
3. Generate an API key in the dashboard. It looks like sk-pool-… and works immediately; the same key and balance also cover the supported Claude, Gemini and Kimi models.
4. Verify the key with the curl below before wiring it into a project. A 200 with real output closes the loop; a 401 means the key or the header name is wrong.

```
curl https://router.apitoken.sale/v1/responses \
  -H "Authorization: Bearer $APITOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.6-terra","input":"Reply with exactly: connected"}'
```

Buying a GPT API key here is three moves — sign up, top up, generate — and the key is active on the very next request: no OpenAI Platform account, no waitlist and no manual review anywhere in the flow. If the curl above returns output text, the whole chain of key, balance and endpoint is proven end to end.

## Which GPT models the key serves

The catalog served to your key is always the live answer at GET https://router.apitoken.sale/v1/models — read it instead of assuming a model name from OpenAI's docs exists here. Today the line covers three GPT-5.6 tiers, two previous-generation models, and the separate GPT Image 2 generation and edit routes:

| Model ID | Tier | Official in / cached / out ($ per 1M) | After 50% discount |
| --- | --- | --- | --- |
| gpt-5.6-sol (alias: gpt-5.6) | Flagship | $4 / $0.40 / $20 | $2 / $0.20 / $10 |
| gpt-5.6-terra | Balanced | $2 / $0.20 / $12 | $1 / $0.10 / $6 |
| gpt-5.6-luna | Fast | $0.20 / $0.02 / $1.20 | $0.10 / $0.01 / $0.60 |
| gpt-5.5 | Previous flagship | $5 / — / $30 | $2.50 / — / $15 |
| gpt-5.4 | Previous balanced | $2.50 / — / $15 | $1.25 / — / $7.50 |

- Sol's temporary official input/cached/cache-write/output rates are $4/$0.40/$5/$20 through 2026-11-21 inclusive, or $2/$0.20/$2.50/$10 here after 50% off. Standard $5 input and $30 output return on 2026-11-22 UTC.
- All three GPT-5.6 tiers share the official 1.05M-token total context window: up to 922K input plus 128K output. They accept text and image input and expose adjustable reasoning effort — none through xhigh, plus max on the GPT-5.6 line.
- Responses and Chat Completions both work with incremental SSE streaming, so existing generate-or-stream loops port without structural changes. Official tool_choice (including required and named tools), parallel_tool_calls including false, and hosted web_search/image_generation are forwarded; Chat Completions maps those hosted tools onto the same Responses tools.
- GPT Image 2 runs on its own generation and edit routes rather than the chat surface; it bills against the same balance. Responses image_generation (on a GPT text model) separately forwards jpeg/webp output_format and partial_images.

[Per-model specs and discounted prices](/models)

[Full GPT pricing breakdown: cache writes and long context](/docs/learn/gpt-api-pricing)

## How the prepaid balance and the 50% discount settle

There is no subscription and no monthly fee. Every request is metered at official OpenAI token rates first, your flat 50% B2C discount is subtracted, and the net amount is drawn from the prepaid balance — so $50 of balance covers $100 of official-rate usage. The dashboard records the settled token usage and exact discounted charge for each request.

- Cached input is priced separately and far cheaper than fresh input ($0.40 vs $4 per 1M on promotional Sol), so a stable prompt prefix across calls is real money.
- Cache writes bill at 125% of normal input; cached reads bill at 10% of input.
- Reasoning tokens appear in output usage and are not charged a second time as a separate leg.

> Above 272K input tokens, long-context rates apply to the whole request — 2× on input and 1.5× on output, before the discount. On promotional Sol, 270K input plus 2K output costs $1.12 official; 273K plus 2K costs $2.244. Split oversized contexts before crossing the boundary.

> When the balance runs out, requests fail with an insufficient-balance error until you top up again — there is no overdraft and no surprise charge to your card.

## Configuring the official SDK and existing clients

Every OpenAI-compatible client needs exactly two values changed: the base URL and the credential. Prompts, streaming code and tool definitions stay as they are — the official SDKs work unchanged:

```
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["APITOKEN_API_KEY"],
    base_url="https://router.apitoken.sale/v1",
)

response = client.responses.create(
    model="gpt-5.6-terra",
    input="Reply with exactly: connected",
)
print(response.output_text)
```

Frameworks that expect the classic Chat Completions shape — older LangChain chains, LiteLLM configs, most open-source chat UIs — work on the same host with the same key and model IDs; only the method name changes to client.chat.completions.create with a messages array.

> Keep the key in a server-side environment variable, never in client-side code or a committed file. A GPT call authenticates with Authorization: Bearer; x-api-key belongs to the Anthropic Messages lane and x-goog-api-key to the native Gemini lane — sending either here returns a 401.

## What this gateway is — and is not

This is an independent OpenAI-compatible gateway with its own account, prepaid balance and supported-model catalog — not the OpenAI Platform. For text, vision, tool calling and hosted web_search/image_generation the surface is complete: model discovery, Responses, Chat Completions, streaming, those hosted tools, and the GPT Image 2 routes. Audio, file, realtime, assistants, batch and fine-tuning endpoints are not available; an app that depends on those is not a candidate for migration.

Errors arrive in the standard OpenAI envelope, so existing error-handling code keeps working. Three status codes cover nearly everything you will hit while integrating:

- 401 — the key is missing or mistyped, or you sent x-api-key instead of Authorization: Bearer. Reproduce with curl outside your app to isolate which half is broken.
- 402 — the prepaid balance needs a top-up; no retry or backoff fixes an empty balance.
- 404 — the model ID is not enabled on your key; check GET https://router.apitoken.sale/v1/models instead of assuming.

[From key to first streaming response: the full quickstart](/docs/learn/openai-api-quickstart)

## Frequently asked questions

### Do I need an OpenAI account to buy this GPT API key?

No. The key, balance and billing come from apiToken.sale; compatible GPT clients only need the custom base URL and the Bearer key. There is no waitlist or manual review.

### Can I pay for a GPT API key with cryptocurrency?

Yes. Checkout accepts bank cards and cryptocurrency, and top-ups are any whole-dollar amount with no fixed bundle. New accounts created with Google or GitHub also start with $5 of platform bonus credit.

### Can one key run both GPT and Claude?

Yes. The same sk-pool key and prepaid balance cover all supported providers; only the endpoint and authorization header change with the protocol — Bearer on the OpenAI-compatible lane, x-api-key on Anthropic Messages.

### How much does GPT-5.6 cost here per 1M tokens?

Through 2026-11-21 inclusive, Sol's temporary official input/cached/cache-write/output rates are $4/$0.40/$5/$20 and settle here at $2/$0.20/$2.50/$10. Standard $5 input and $30 output return on 2026-11-22 UTC. Terra remains $2/$12 and Luna $0.20/$1.20 officially.

### Is this the OpenAI Platform?

No. It is an independent OpenAI-compatible gateway with its own account, prepaid balance and supported-model catalog. Responses, Chat Completions, streaming, hosted web_search/image_generation and GPT Image 2 are served; audio, realtime, assistants, batch and fine-tuning are not.

---
Get a key: https://apitoken.sale/register
More guides: https://apitoken.sale/docs/learn