AI Model Pricing — Pay Per Prompt
Pay only for what you use.
How Pricing Works
Quick Overview
Pricing is based on tokens (roughly 3 words ≈ 4 tokens). You pay for input tokens (what you send) and output tokens (what the AI responds).
“Explain the concept of artificial intelligence in simple terms. Pretend that I'm a 5-year old that doesn't have a very long attention span, so keep it limited to roughly 150 words.”
“Okay! Imagine you have a really smart robot friend. This robot can learn things just like you do...”
Tip: Click the Input/Output token badges above to edit.
Looking for our subscription? View subscription
Text Model Pricing
No models found
Try searching with a different term or clear your search
* Rates shown per 1M tokens. Long-context models list a separate `200k+` or `256k+` input-tier row. Cache line shows cached input read/write pricing.
Image Model Pricing
Video Model Pricing
Important Note
Prices shown are minimum costs (cheapest resolution/duration). Actual costs vary based on settings.
Audio Model Pricing
Text-to-Speech
* TTS rates shown per 1k characters unless marked fixed.
Speech-to-Text
* STT rates shown per billing minute. Some models round each request up to the next full minute.
Embedding Model Pricing
About Embedding Pricing
At-cost pricing per 1M tokens. No markup or additional discounts applied.
* Rates shown per 1M tokens.
Need Volume Pricing?
If you're a large user of our platform or API, we'd love to work with you. Contact us for custom pricing and dedicated support.
Explore Pricing Resources
Learn more about pricing, model comparisons, and how to get started with NanoGPT.