Model Pricing Details
  • The price shown above the chart is the INPUT price (per 1M tokens).
  • The OUTPUT price is typically 3-5 times higher than the INPUT price.
  • The OUTPUT price is displayed when selecting each model on the chart.
  • OpenAI models are priced 50% lower when the Batch API is used.
  • The Google Gemini API offers a "free tier" with lower rate limits for testing purposes.
  • The price shown for Gemini on the above chart is "Pay-as-you-go" for prompts up to 128k tokens (fees double for prompts longer than 128k tokens).
  • Some models also charge for context caching (typically 25% lower than INPUT pricing).

What does the rank mean?

The rank is from the LMSYS Chatbot Arena Leaderboard. It has over 1,000,000 human pairwise comparisons to rank LLMs with the Bradley-Terry model and display the model ratings in Elo-scale. You can find more details in their paper.

Better value models are found in the top-left (high rank and low cost).

Where are the prices from?

I sourced the prices from the websites of the companies who made them and host them. There are other companies that host the models, and I may list them in an update.

You can find the prices here:

I don't see [INSERT MODEL NAME], where is it?

There are a few reasons a model might not show up:

See an issue or error?

Send me an email at info@llmcompare.net