Prepaid, OpenAI-compatible API key

Pay-as-you-go API for GPT, DeepSeek, Kimi and Fish Audio.

One prepaid key for 16 models from 4 providers. Add credit from $5 and pay each provider’s list price per token.

Model prices: The provider’s list price. Nothing is added per request.
Top-up fee: 5%, minimum $0.79, shown before you pay: $25 of credit costs $26.25.
Subscription: None. Paid credit never expires.

Add credit See all prices

OpenAI-compatible. Base URL https://api.airuntime.dev/v1; the official SDKs work unchanged.
Refunds. Unused paid credit within 14 days, fee included. Terms
Operated by Grid Heap, Inc., Dover, Delaware, USA. Contact

Example: what one request costs

POST /v1/chat/completions

System You are the support assistant for a small SaaS. Answer in plain English… 1,024 tokens, cached

User Explain prompt caching in two sentences.

Reply Prompt caching keeps the start of a prompt you send often, like a long system message. When a request begins the same way, the provider reads those tokens from its cache and bills them at the cached rate.

Cost of the example request by token type
Tokens	Count	Rate	Cost
Input	22	$0.15/1M	$0.000003
Cached input	1,024	$0.003/1M	$0.000003
Output	42	$0.60/1M	$0.000025
Total for this request			$0.000032

$5 of credit covers about 158,368 requests like this with DeepSeek V4.1 Flash. Token counts are illustrative; rates are the provider’s list prices.

Price list

API prices for all 16 models

Text models are priced per 1M tokens: input, cached input (when the provider’s prompt cache hits) and output. Image, speech, transcription and embedding models are priced per unit. These are the providers’ list prices; the 5% fee (minimum $0.79) applies only when you add credit.

Pricing, by provider

16 models from 4 providers

Text models

US dollars per 1M tokens.

Text models and prices per 1M tokens
Model	Context	Input$/1M	Cached$/1M	Output$/1M
DeepSeek V4.1 Flash `deepseek-flash`DeepSeek2× at peak	1M	$0.15	$0.003	$0.60
GPT-6 Astra `gpt-6-astra`OpenAI	1.05M	$10.00	$1.00	$50.00
GPT-6.1 Sol `gpt-6-1-sol`OpenAI	1.05M	$2.00	$0.10	$10.00
GPT-6.1 Terra `gpt-6-1-terra`OpenAI	1.05M	$1.00	$0.05	$5.00
GPT-6 Luna `gpt-6-luna`OpenAI	1.05M	$0.10	$0.01	$0.50
Kimi K3.1 Pro `kimi-k3-1-pro`Moonshot AI	1M	$1.92	$0.192	$10.04
Kimi K3.1 Pro Highspeed `kimi-k3-1-pro-highspeed`Moonshot AI	1M	$3.84	$0.192	$20.08

Images models

2 models, priced per unit.

Images models and prices
Model	Endpoint	Price
GPT Image 2.5 Sunburst `gpt-image-2-5-sunburst`OpenAI	`/v1/images/generations`	$0.006 per image
GPT Image 2.5 Flare `gpt-image-2-5-flare`OpenAI	`/v1/images/generations`	$0.006 per image

Speech models

3 models, priced per unit.

Speech models and prices
Model	Endpoint	Price
GPT-4o Mini TTS `gpt-4o-mini-tts`OpenAI	`/v1/audio/speech`	$0.015 per 1K characters
Fish S2.1 Pro `fish-s2-1-pro`Fish Audio	`/v1/audio/speech`	$0.015 per 1K bytes
Fish S2.1 Pro Free `fish-s2-1-pro-free`Fish Audio	`/v1/audio/speech`	Free per 1K bytes

Transcription models

3 models, priced per unit.

Transcription models and prices
Model	Endpoint	Price
Transcription Fast `gpt-4o-mini-transcribe`OpenAI	`/v1/audio/transcriptions`	$1.25 per 1M audio tokens
Transcription Accurate `gpt-4o-transcribe`OpenAI	`/v1/audio/transcriptions`	$2.50 per 1M audio tokens
Fish Transcribe 1 Pro `fish-transcribe-1-pro`Fish Audio	`/v1/audio/transcriptions`	$0.006 per minute

Embeddings models

One model, priced per unit.

Embeddings models and prices
Model	Endpoint	Price
Text Embeddings `text-embedding-3-small`OpenAI	`/v1/embeddings`	$0.02 per 1M tokens

How to start

Get an API key in three steps

Add credit

Sign in with a Gridheap Account and add any amount from $5 to $500, in $5 steps. A 5% fee (minimum $0.79) is added at checkout, so $25 of credit costs $26.25.
Copy your API key

Your rk_ key is shown once, right after payment. Save it in your password manager; it works with every model on this site.
Call any model

Point an OpenAI SDK at api.airuntime.dev/v1, use your key, and set model to the id from the price list. Each request costs the provider’s list price.

API

OpenAI-compatible API: keep your SDK, change the base URL

Use the OpenAI SDK for Python or Node, or any client that speaks the OpenAI API. Set the base URL, use your key, and set model to the id from the price list.

Base URL: https://api.airuntime.dev/v1
Auth: Authorization: Bearer rk_…
Model: deepseek-flash or any id in the price list
Endpoints: /v1/chat/completions/v1/images/generations/v1/audio/speech/v1/audio/transcriptions/v1/embeddings

curl

curl https://api.airuntime.dev/v1/chat/completions \
  -H "Authorization: Bearer $TOKENS_PAY_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-flash",
    "messages": [{"role": "user", "content": "Say hello in five words."}]
  }'

Python

from openai import OpenAI
import os

client = OpenAI(base_url="https://api.airuntime.dev/v1", api_key=os.environ["TOKENS_PAY_KEY"])
reply = client.chat.completions.create(
    model="deepseek-flash",
    messages=[{"role": "user", "content": "Say hello in five words."}],
)
print(reply.choices[0].message.content)

Node

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.airuntime.dev/v1", apiKey: process.env.TOKENS_PAY_KEY });
const reply = await client.chat.completions.create({
  model: "deepseek-flash",
  messages: [{ role: "user", content: "Say hello in five words." }],
});
console.log(reply.choices[0].message.content);

Zero data retention

We don’t store your prompts or outputs.

AI Runtime, the API your key calls, holds each request in memory while it runs and keeps only the bill. Tokens.Pay never sees your requests at all.

Zero data retention, exactly Security

Prompts are not stored: AI Runtime, the gateway your key calls, does not write prompts, messages, files, images, audio or embedding inputs to disk, a database, logs or analytics. They exist in memory only while the request runs.
Outputs are not stored: Completions, generated images, speech and transcripts go back to you and are not kept.
No training, no human review: Your content is not used to train or evaluate models, and no person at Gridheap reads it.
Providers see us, not you: AI Runtime rebuilds each request field by field before it goes to the model’s provider, with no name, email, account or key of yours in it. To the provider, the customer is AI Runtime.

Dashboard

Track spend, usage and cache hits per model

Your Dashboard shows your balance, spend by day and by model, your latest calls and your cache hit rate, as soon as the provider reports them.

Dashboard previewExample data

Spent, 30 days: $28.36
Cache hit rate: 70%; of input tokens, at the cache price
Tokens: 74.9Min; 9.4M out

Example spend by model, last 30 days
Model	Calls	Cache hits	Cost
DeepSeek V4.1 Flash	18,420	73%	$7.02
GPT-6.1 Terra	2,310	52%	$11.71
Kimi K3.1 Pro	640	59%	$9.64

Pricing

Add prepaid credit

A $25 top-up costs $26.25: $25.00 of credit and a $1.25 fee. Requests then cost exactly what the provider charges.

Model prices: Each provider’s list price, per token or per unit. Cached input is billed at the provider’s lower cache rate.
Service fee: 5% on each top-up, minimum $0.79, shown before you pay. Nothing is added per request.
Credit: Paid credit never expires. No subscription, no automatic recharge.
Welcome credit: $1 when you sign up, and $5 more with your first top-up. Promotional credit is not refundable.
Refunds: Unused paid credit can be refunded within 14 days of the top-up, fee included. Refund terms
Operator: Grid Heap, Inc., 1111B S Governors Ave Ste 51059, Dover, DE 19904, USA.

Credit: $25.00
Top-up fee (5%): $1.25
You pay: $26.25

$25 of credit buys about 42M output tokens with DeepSeek V4.1 Flash

Estimate with

New accounts get $1, plus $5 with the first top-up.

You’ll sign in first. Nothing is charged until you confirm on the payment page.

FAQ

Frequently asked questions

Something else? Write to support. A person answers.

Which SDKs and tools work with my key?

Anything that speaks the OpenAI API. Set the base URL to https://api.airuntime.dev/v1 and use your key as the API key. The official OpenAI SDKs for Python and Node work unchanged.

Is there a markup on model prices?

No. Each request costs the provider’s list price. The only extra is the 5% fee on each top-up (minimum $0.79), which you see before paying.

Does my credit expire?

Paid credit never expires. Use it at your own pace, on any model in the list.

What happens if I lose my API key?

Keys are shown once. On your Profile page you can turn a lost key off and get a new one. Your balance stays the same.

How does prompt caching affect the price?

When a provider serves part of your prompt from its cache, those input tokens are billed at the provider’s cache rate, often a tenth of the normal input price or less. The price list shows both rates, and your Dashboard shows your cache hit rate.

Why does DeepSeek cost more at some hours?

DeepSeek charges double on weekdays from 01:00 to 04:00 and from 06:00 to 10:00 UTC. We pass its price through unchanged, so the list shows the rate that applies right now.

Do you store my prompts or outputs?

No. AI Runtime, the API your key calls, holds each request in memory only while it runs and keeps just the billing record: id, time, model, token counts, cost and status. Tokens.Pay never sees your requests. The model provider processes them under its own data terms. The details are on the zero data retention page.

Who operates Tokens.Pay?

Grid Heap, Inc., a Delaware corporation, 1111B S Governors Ave Ste 51059, Dover, DE 19904, USA. Sign-in is a Gridheap Account; payments are taken on Stripe’s or PayPal’s own page.

Can I get a refund?

Yes, for unused paid credit within 14 days of the top-up, including its fee. Write to support with the payment date.