API Pricing

This page outlines the API cost structures and billing process for our two primary service offerings: Clarity API and Portkey. For a comprehensive comparison of the features and intended use cases for each service, please visit the API page.

Understanding the Costs

Tokens and Cost Estimation

AI models process text in pieces called tokens. For English text, one token is approximately 4 characters or ¾ of a word. Your total cost is determined by two factors: the specific AI model used and the number of tokens in your input (the “prompt”) plus the number of tokens in the model’s output (the “completion”).

Examples:

  • Submitting a one-sentence prompt (15 words) with a one-paragraph reply (100 words) costs approximately $0.002 using GPT-5.2 through Portkey.
  • Submitting a longer document (4,000 words) with a four-paragraph summary (600 words) costs approximately $0.03 using Gemini 2.5 Pro through Clarity API.

Billing via Chart of Accounts (COA)

Our billing cycle is designed to be simple and transparent.

  1. Request API Access: When you submit an API Use Case Request Form, you will provide a COA for billing of your use case. The COA will be validated before granting you access to either Portkey or Clarity API.
  2. Usage: Usage is tracked for your API keys.
  3. Monthly Billing: At the end of each month, you will be charged directly to the designated COA.

To change the COA used for billing, please contact us at ai.platforms@yale.edu before the next billing cycle begins.

Portkey API Pricing

Portkey offers direct, cost-effective access to a range of AI models. When using Portkey, you are charged only for the underlying AI model’s usage costs, with no additional service fees. Portkey is an ideal choice for use cases where direct model access and cost efficiency are the primary requirements.

Cached Prompts: Prompt caching helps reduce latency and costs by caching parts of your prompt for reuse instead of sending the same prompt multiple times. For a detailed overview of which AI models support caching, please refer to Portkey’s website.

Usage Monitoring: You can track your usage metrics in real-time directly within the Portkey platform. It also allows you to set rate limits and budget alerts for your API keys.

Model Costs

Portkey tracks the estimated cost for model usage that you can view at any time in Portkey. The table below outlines the differences between global inferencing pricing and US region pricing. Please note that prices are subject to change.

Category Global Inferencing Pricing US Region Pricing
Description
  • Models leveraging global distribution of API requests
  • Distributions across multiple regions
  • Regions may be outside the US
  • Models leveraging US distribution of API requests
  • Will not distribute outside of US regions
Costs Portkey Model Pricing
  • Portkey Model Pricing
  • 10% higher than global inferencing (not reflected in Portkey’s table)
  • Set budget limits in Portkey with 10% increase in mind
Data Risk Classification Low-risk data
  • Moderate to high-risk data
  • Yale’s Minimum Security Standards only allow US distribution for these data classifications

Clarity API Pricing

Clarity is Yale’s secure AI platform, and its API key offering is designed for use cases that require enhanced security and advanced features. Its pricing consists of the underlying model cost plus a service fee to cover these premium capabilities.

Usage Monitoring: As the owner of the API agent, you will be granted access to a Power BI dashboard to track usage and spending. This dashboard is refreshed weekly.

Model Costs including Service Fees (per 1M Tokens)

Please note: Prices are subject to change based on the costs of the underlying AI models. The costs outlined in the table below are current as of May 2026.

AI Model Costs for Clarity API
Model Name Input Cost Output Cost 
GPT-5.2  $3.50 $28.00 
GPT-5 mini  $0.50 $4.00
o3  $4.00 $16.00
Claude Sonnet 4.6 $6.00  $30.00 
Gemini 2.5 Flash  $0.60  $5.00 
Gemini 2.5 Pro  $2.50  $20.00