Understanding Token Costs

Chibi AI is "bring your own AI." This means you connect your OpenRouter account to Chibi and then pay token costs directly with them, separate from Chibi.

What is a token?

Here's how OpenAI describes tokens:

"Tokens can be thought of as pieces of words. Before the API processes the prompts, the input is broken down into tokens. These tokens are not cut up exactly where the words start or end - tokens can include trailing spaces and even sub-words." -OpenAI

Different models have different costs

We support major AI models from providers such as Anthropic, OpenAI, AI21, Cohere, Together AI, Fireworks, etc. all through your OpenRouter account. Each model has different costs.

Here's some perspective:

  • 1M tokens with a Llama 3.1 8B parameter model is just $0.06!

  • A larger model like Mistral Large is ~$6 for 1M tokens.

  • OpenAI GPT-4o is about $10 for 1M tokens, while GPT-4o Mini is just $0.38 for 1M tokens.

  • You can think of one token as roughly 3-4 characters of text on average.

  • That’s 3-4 million characters of text for the prices above.

  • The average character count of a Harry Potter book is something like 800,000.

That means an entire Harry Potter book is less than $0.08 of tokens if you use GPT-4o Mini!

Pricing pages

Check OpenRouter for the models we support, including their prices.

Was this helpful?