EN
EnglishРусскийTürkçeEspañol
Log inGet API key
NewGPT-6 models are live

One API key for GPT, Codex, DeepSeek and GLM

Prepaid and OpenAI-compatible. From $0.02 per 1M tokens, top up from $5, no subscription.

Try with $0.10 free creditRead the docs
api.model.sale/v1
curl https://api.model.sale/v1/responses \
  -H "Authorization: Bearer $MODEL_SALE_KEY" \
  -d '{"model":"gpt-5.5","input":"Hello"}'
Modelper 1M
gpt-6-astra$1.13
deepseek-flash$0.03
glm-5.2$0.15
gpt-6-luna$0.08
deepseek-v4-pro$0.02
glm-5.3$0.15
gpt-6-sol$0.38
All prices →
200 OKgpt-6-astra1,284 tokenscharged $0.001445example
13models live now
$0.02from, per 1M tokens
$5minimum top-up
0subscriptions or seat fees
Works withCodex CLIOpenAI SDKCursorVS CodeZedOpenCodecurl

Live now

All models →
ChatGPT / Codex
gpt-6-astra
$1.125 / 1M tokensResponses
GLM
glm-5.2
$0.15 / 1M tokensChat Completions
ChatGPT / Codex
gpt-6-luna
$0.08 / 1M tokensResponses
GLM
glm-5.3
$0.15 / 1M tokensChat Completions
ChatGPT / Codex
gpt-6-sol
$0.38 / 1M tokensResponses
ChatGPT / Codex
gpt-6.1-sol
$0.40 / 1M tokensResponses

Why Model.sale

Built for agents and production code.

Drop-in compatibleResponses and Chat Completions for the OpenAI SDKs, Codex CLI, Cursor and any HTTP client. Change two lines.
Prepaid, never surprisedTop up from $5. Each request holds a small amount and settles to the exact tokens used — no overages, no subscription.
Private by defaultWe keep billing metadata only — never your prompts or model outputs.
Per-key controlsLimit models, spend per day and month, RPM, TPM and IPs for every key.
Live health checksEvery model is probed continuously; unhealthy routes stop accepting traffic before you notice.
Usage you can auditInput, cached, output and reasoning tokens with the exact charge for every request in your dashboard.

From the blog

Guides for choosing and using model APIs.

ChatGPT API, Codex or GLM? Choosing the right endpointA practical guide to ChatGPT-compatible Responses, Codex clients and GLM Chat Completions with one prepaid API key.Baidu Qianfan vs DeepSeek vs GLM: choosing a Chinese model APIA practical comparison of Chinese model APIs based on compatibility, billing, reliability and commercial-use checks.The cheapest ChatGPT API is not the one with the lowest priceA practical way to compare ChatGPT-compatible APIs using the same token mix, streaming test and final settled charge.

Get started

Three steps to your first request.

  1. 1Create an accountSign in with Google, Telegram or email. No card needed to look around.
  2. 2Top up from $5Card, PayPal, Apple Pay or crypto. Estimate spend with the cost calculator.
  3. 3Create a key and callSet OPENAI_BASE_URL and your key — follow the setup guides for Codex, Cursor and SDKs.

FAQ

Questions, answered.

How can I pay?

Top up from $5 by card, PayPal or Apple Pay through Paddle, or with crypto through Crypto Pay. There is no subscription and the balance does not expire.

Which tools work?

Any OpenAI-compatible client: the OpenAI SDKs, Codex CLI, Cursor, OpenCode, Zed or plain HTTP. See the setup guides.

What if a request fails?

A balance hold is placed before the request and released if the model fails. You pay only for usage the model actually reports.

What do I pay for?

Only the tokens a completed request actually used: input and output at the model price, cached input at 40% of it. Input, output, cached and reasoning fields remain visible separately; the unused reservation is released.

How do I start?

Create an account, generate an API key, add balance and point your existing SDK at api.model.sale.

Are prompts stored?

Request and response bodies are not stored by default. We retain billing and operational metadata.

Your first request in under a minute.

Create a key, top up from $5 and swap the base URL. No subscription, no hidden fees.

Get your API keyRead the docs