# Tokens.Pay

> Prepaid credit for AI model APIs. Buy credit once (from $5, plus a 5% top-up fee with a $0.79 minimum), get one OpenAI-compatible API key, and call 16 models from OpenAI, Fish Audio, Moonshot AI and DeepSeek at the providers' list prices.

- API base URL: https://api.airuntime.dev/v1 (OpenAI-compatible; send the key as `Authorization: Bearer <key>`)
- Every model and price: https://tokens.pay/models/
- Buy credit: https://tokens.pay/#pay (sign-in with a Gridheap Account, card via Stripe or PayPal)
- Paid credit never expires; no subscription. The API key is shown once after payment.
- Pricing, refunds and what happens at $0: https://tokens.pay/pricing/
- Data handling: AI Runtime does not store prompts or outputs (zero data retention); details and exceptions at https://tokens.pay/privacy/
- Support: support@gridheap.com
- Terms (including refunds and acceptable use): https://www.gridheap.com/legal/terms#tokens-pay

## Docs

- [Quickstart](https://tokens.pay/docs/): Get a prepaid Tokens.Pay API key, set the base URL https://api.airuntime.dev/v1 in any OpenAI SDK and make your first chat completion.
- [Authentication](https://tokens.pay/docs/authentication/): Authenticate Tokens.Pay API requests with your rk_ key as a Bearer token. Key format, where to keep it, and the 401 and 403 errors you can get.
- [Models and endpoints](https://tokens.pay/docs/models/): The model id to send for each of the 16 models on Tokens.Pay, the endpoint each one answers on, and how to list the models your key can call.
- [Streaming](https://tokens.pay/docs/streaming/): Stream chat completions as server-sent events: stream: true, usage in the last chunk with include_usage, the [DONE] marker and interrupted streams.
- [Reasoning](https://tokens.pay/docs/reasoning/): Set reasoning_effort on reasoning models, read reasoning_content and usage.completion_tokens_details.reasoning_tokens, and how thinking tokens are billed.
- [Errors and rate limits](https://tokens.pay/docs/errors/): Error format and codes from the Tokens.Pay API: invalid requests, keys, balance, rate limits and provider failures, and when and how to retry.
- [Billing and usage](https://tokens.pay/docs/billing/): How Tokens.Pay bills API calls: list price per token, cached input, reasoning tokens, the cost on every response, failed requests and your Dashboard usage.

## Models

- [DeepSeek V4.1 Flash](https://tokens.pay/models/deepseek-flash/): model id `deepseek-flash`, /v1/chat/completions, $0.15 input, $0.003 cached input, $0.60 output per 1M tokens off-peak (2× at DeepSeek peak), 1M context
- [GPT-6 Astra](https://tokens.pay/models/gpt-6-astra/): model id `gpt-6-astra`, /v1/chat/completions, $10.00 input, $1.00 cached input, $50.00 output per 1M tokens, 1.05M context
- [GPT-6.1 Sol](https://tokens.pay/models/gpt-6-1-sol/): model id `gpt-6-1-sol`, /v1/chat/completions, $2.00 input, $0.10 cached input, $10.00 output per 1M tokens, 1.05M context
- [GPT-6.1 Terra](https://tokens.pay/models/gpt-6-1-terra/): model id `gpt-6-1-terra`, /v1/chat/completions, $1.00 input, $0.05 cached input, $5.00 output per 1M tokens, 1.05M context
- [GPT-6 Luna](https://tokens.pay/models/gpt-6-luna/): model id `gpt-6-luna`, /v1/chat/completions, $0.10 input, $0.01 cached input, $0.50 output per 1M tokens, 1.05M context
- [Kimi K3.1 Pro](https://tokens.pay/models/kimi-k3-1-pro/): model id `kimi-k3-1-pro`, /v1/chat/completions, $1.92 input, $0.192 cached input, $10.04 output per 1M tokens, 1M context
- [Kimi K3.1 Pro Highspeed](https://tokens.pay/models/kimi-k3-1-pro-highspeed/): model id `kimi-k3-1-pro-highspeed`, /v1/chat/completions, $3.84 input, $0.192 cached input, $20.08 output per 1M tokens, 1M context
- [GPT Image 2.5 Sunburst](https://tokens.pay/models/gpt-image-2-5-sunburst/): model id `gpt-image-2-5-sunburst`, /v1/images/generations, $0.006 per image
- [GPT Image 2.5 Flare](https://tokens.pay/models/gpt-image-2-5-flare/): model id `gpt-image-2-5-flare`, /v1/images/generations, $0.006 per image
- [GPT-4o Mini TTS](https://tokens.pay/models/gpt-4o-mini-tts/): model id `gpt-4o-mini-tts`, /v1/audio/speech, $0.015 per 1K characters
- [Fish S2.1 Pro](https://tokens.pay/models/fish-s2-1-pro/): model id `fish-s2-1-pro`, /v1/audio/speech, $0.015 per 1K bytes
- [Fish S2.1 Pro Free](https://tokens.pay/models/fish-s2-1-pro-free/): model id `fish-s2-1-pro-free`, /v1/audio/speech, free
- [Transcription Fast](https://tokens.pay/models/gpt-4o-mini-transcribe/): model id `gpt-4o-mini-transcribe`, /v1/audio/transcriptions, $1.25 per 1M audio tokens
- [Transcription Accurate](https://tokens.pay/models/gpt-4o-transcribe/): model id `gpt-4o-transcribe`, /v1/audio/transcriptions, $2.50 per 1M audio tokens
- [Fish Transcribe 1 Pro](https://tokens.pay/models/fish-transcribe-1-pro/): model id `fish-transcribe-1-pro`, /v1/audio/transcriptions, $0.006 per minute
- [Text Embeddings](https://tokens.pay/models/text-embedding-3-small/): model id `text-embedding-3-small`, /v1/embeddings, $0.02 per 1M tokens

## Pages

- [Pay-as-you-go API for GPT, DeepSeek and Kimi · Tokens.Pay](https://tokens.pay/): Prepaid, OpenAI-compatible API key for 16 AI models: GPT, DeepSeek and Kimi. List prices per token, 5% top-up fee (minimum $0.79), no subscription.
- [AI model API prices: GPT, DeepSeek and Kimi](https://tokens.pay/models/): API prices for 16 AI models from 4 providers, per 1M tokens or per unit, at list price. One prepaid OpenAI-compatible key.
- [OpenAI API pricing: GPT models per 1M tokens](https://tokens.pay/models/openai/): GPT API pricing from $0.10 per 1M input tokens: 10 OpenAI models at list price. One prepaid OpenAI-compatible key, from $5.
- [Fish Audio API pricing](https://tokens.pay/models/fish-audio/): Fish Audio API pricing: Fish S2.1 Pro, Fish S2.1 Pro Free and Fish Transcribe 1 Pro at list price. One prepaid OpenAI-compatible key, from $5.
- [Kimi API pricing (Moonshot AI) per 1M tokens](https://tokens.pay/models/moonshot-ai/): Kimi API pricing from $1.92 per 1M input tokens: Kimi K3.1 Pro and Kimi K3.1 Pro Highspeed at list price. One prepaid OpenAI-compatible key, from $5.
- [DeepSeek API pricing per 1M tokens](https://tokens.pay/models/deepseek/): DeepSeek API pricing: DeepSeek V4.1 Flash $0.15 input, $0.60 output per 1M tokens. Prepaid OpenAI-compatible key, from $5.
- [LLM API pricing per 1M tokens](https://tokens.pay/models/text-generation/): LLM API pricing for 7 models from DeepSeek, OpenAI and Moonshot AI at list price. One prepaid OpenAI-compatible key, from $5.
- [Reasoning model API pricing per 1M tokens](https://tokens.pay/models/reasoning/): Reasoning model API pricing for 6 models from DeepSeek, OpenAI and Moonshot AI at list price. One prepaid OpenAI-compatible key, from $5.
- [Vision model API pricing: image input](https://tokens.pay/models/vision/): Vision model API pricing for 7 models from DeepSeek, OpenAI and Moonshot AI at list price. One prepaid OpenAI-compatible key, from $5.
- [Image generation API pricing per image](https://tokens.pay/models/image-generation/): Image generation API pricing: GPT Image 2.5 Sunburst $0.006 per image; GPT Image 2.5 Flare $0.006 per image. One prepaid OpenAI-compatible key.
- [Text-to-speech API pricing](https://tokens.pay/models/text-to-speech/): Text-to-speech API pricing for 3 models from OpenAI and Fish Audio at list price. One prepaid OpenAI-compatible key, from $5.
- [Speech-to-text API pricing](https://tokens.pay/models/speech-to-text/): Speech-to-text API pricing for 3 models from OpenAI and Fish Audio at list price. One prepaid OpenAI-compatible key, from $5.
- [Embeddings API pricing per 1M tokens](https://tokens.pay/models/embeddings/): Embeddings API pricing: Text Embeddings $0.02 per 1M tokens. One prepaid OpenAI-compatible key.
- [AI API pricing: prepaid credit, list prices](https://tokens.pay/pricing/): Prepaid AI API credit from $5 to $500, a 5% top-up fee (minimum $0.79) and the providers’ list price per token for 16 models. No subscription.
- [Security: API keys, encryption, payments](https://tokens.pay/security/): How Tokens.Pay protects your API key (SHA-256, shown once), encrypts data in transit and at rest, takes payments via Stripe and PayPal, and where it runs.
- [Zero data retention: prompts not stored](https://tokens.pay/privacy/): Tokens.Pay and AI Runtime don’t store your prompts or outputs. The exceptions, the billing record we keep and which providers process requests where.