103 models · language, image and video

One API key for 103 AI models

Call Claude, GPT, Gemini, DeepSeek, Seedance, Kling and more with the same key and the same balance. Pay for each call you make, nothing per month.

See prices

Sign-up and top-ups open soon.

models at launch
103
API key for all of them
1
minimum top-up
$10
for failed calls
$0

Why KeelRouter

  • One key, one balance

    No separate accounts and invoices for each lab. Top up once and spend it on any model.

  • Below list price on many models

    Several popular models cost less here than at the lab that makes them. The table below shows both prices.

  • Drop-in for existing SDKs

    Chat models answer on OpenAI- and Anthropic-compatible endpoints. Change the base URL and the key, keep your code.

  • Every call has a receipt

    Each request gets an ID. Look it up to see the model, the status and the exact amount charged.

  • Fallback routes

    Popular models run on more than one upstream route. If one is down or rate-limited, the call moves to the next.

  • Credits never expire

    Top-ups stay in your balance until you use them. New accounts start with $0.10 to try a call.

Compared with official prices

Official prices from each lab’s own price list, checked 2026-10-24.

ModelTypeKeelRouterOfficial list price
Gemini 4 ArgonGoogleLanguage$1 / $5.20 per 1M tokens$2 / $10
Gemini 3.8 FlashGoogleLanguage$0.29 / $1.40 per 1M tokens$0.75 / $3.75
Claude Sonnet 5.5 *AnthropicLanguage$0.68 / $3.40 per 1M tokens$2 / $10
Claude Opus 5.5 *AnthropicLanguage$1.30 / $6.80 per 1M tokens$4 / $20
Nano Banana 2.1GoogleImage$0.022 per 1K image$0.0336
Veo 3.1GoogleVideo$1.41 per 8-second 1080p clip$3.20

* Claude models run on a Claude Code–compatible route. The provider adds a short system prompt (about 30 tokens) before yours; it counts toward input tokens, and answers can differ slightly from Anthropic’s own API.

Popular models

ModelTypePrice
Gemini 4 ArgonGoogleLanguage$1 / $5.20 per 1M tokens
Gemini 3.8 FlashGoogleLanguage$0.29 / $1.40 per 1M tokens
GPT-6.1 SolOpenAILanguage$0.24 / $1.20 per 1M tokens
Claude Sonnet 5.5 *AnthropicLanguage$0.68 / $3.40 per 1M tokens
Claude Opus 5.5 *AnthropicLanguage$1.30 / $6.80 per 1M tokens
Nano Banana 2.1GoogleImage$0.022 per 1K image
GPT Image 2.5 FlareOpenAIImage$0.015 per 1K image (medium)
Seedream 5.0 FlashByteDanceImage$0.021 per image
MiniMax Hailuo H3MiniMaxVideo$0.052 per second (768p)
Seedance 2.0ByteDanceVideo$0.102 per second (480p)
Kling 4.0KuaishouVideo$0.157 per second (720p)
Veo 3.1GoogleVideo$1.41 per 8-second 1080p clip

* Claude models run on a Claude Code–compatible route. The provider adds a short system prompt (about 30 tokens) before yours; it counts toward input tokens, and answers can differ slightly from Anthropic’s own API.

How billing works

Good to know

  • Prices follow the upstream labs and can change; the price on the model page at the time of the call is what you pay.
  • Result files are deleted after 30 days. Download what you want to keep.
  • Each model keeps its own safety filters. Requests that break the usage policy are refused and not charged.
  • No relaxed-moderation or adult models.
  • Claude models run on a Claude Code–compatible route. The provider adds a short system prompt (about 30 tokens) before yours; it counts toward input tokens, and answers can differ slightly from Anthropic’s own API.

Get started in three steps

  1. 1

    Sign in

    Sign in with Google. No forms, no sales call.

  2. 2

    Top up and create a key

    Add $10 or more by card or Alipay, then create an API key in the console.

  3. 3

    Make your first call

    Point your OpenAI or Anthropic SDK at our base URL, or send a generation request with curl.

FAQ

Is it open yet?
Not yet. The site lists the service and prices; sign-up, top-ups and the API open soon.
How am I billed?
From a prepaid USD balance. Each call is charged by the model's unit: tokens for language models, images for image models, seconds or clips for video. Top-ups start at $10 and larger ones get bonus credit.
Are failed calls charged?
No. If a call fails or is refused, nothing is taken from your balance. For async jobs the estimated cost is held when the job starts and released if it fails.
Do I need to change my code?
For chat models, usually only the base URL and the key. Image and video models use one async endpoint per type: create a task, then poll it or receive a callback.
Which payment methods do you take?
Cards and Alipay, through Stripe. Card payments get an invoice.
Do credits expire?
No. Your balance stays until you use it.
Can I get a refund?
Duplicate charges and billing errors are refunded. Credits already spent are not. See the refund policy for details.
Is Claude here the same as Anthropic’s API?
Claude models run on a Claude Code–compatible route. The provider adds a short system prompt (about 30 tokens) before yours; it counts toward input tokens, and answers can differ slightly from Anthropic’s own API.
Are you affiliated with the model makers?
No. KeelRouter is an independent service. Model names belong to their owners and are used only to say which model a request runs on.

One key for every model you use

Questions? Write to support@keelrouter.com.