Clone OpenAI pricing

Manually managing per-token pricing can be complex, but with Flexprice, you can automate and scale your billing effortlessly. This guide walks you through setting up OpenAI’s O1 pricing model using a package-based pricing approach. This method ensures clarity in billing, making it ideal for AI APIs, generative models, and machine learning services.

Use Cases

• AI APIs (LLMs like OpenAI, Anthropic, Mistral)

• Machine learning inference services

• Text-to-Speech or Speech-to-Text APIs

OpenAI charges users based on the number of input, output and cached tokens processed by their models. You can view the official pricing details here: OpenAI Pricing.

The O1 model has the following pricing structure:

Token Type	Price per Million Tokens
Input Tokens	$15.00 per million tokens
Output Tokens	$60.00 per million tokens
Cached Input Tokens	$7.50 per million tokens

For example, if a user processes:

• 5 million input tokens → $75.00 ($15.00 x 5)

• 2 million output tokens → $120.00 ($60.00 x 2)

• 1 million cached input tokens → $7.50 ($7.50 x 1)

• Total Cost = $202.50

Now, let’s configure the pricing for the O1 model using Flexprice.

Configuring Pricing of o1 model in Flexprice

1. Create Metered Features for Token Usage Since token usage is metered, we first define three separate Metered Features in Flexprice for input tokens, output tokens, and cached input tokens.

   Feature Name	Feature Type	Aggregation Method	Key	Filters
   Input Tokens	Metered	SUM	model_name	model: OpenAI O1, prompt_type: input
   Output Tokens	Metered	SUM	model_name	model: OpenAI O1, prompt_type: output
   Cached Input Tokens	Metered	SUM	model_name	model: OpenAI O1, prompt_type: cached_input

2. Create a Plan with Package-Based Pricing Once the metered features are created, we define a Plan that charges users per million tokens rather than per individual token.

   Metered Feature	Billing Model	Charges
   Input Tokens	Package Charge	$15.00 per million tokens
   Output Tokens	Package Charge	$60.00 per million tokens
   Cached Input Tokens	Package Charge	$7.50 per million tokens

Now, whenever a customer purchases this plan and starts using it, they will:

• See real-time usage events for token consumption.

• Get a dynamically generated proposed invoice based on their usage.

• Have full transparency in billing, ensuring clarity on costs.

This process ensures that AI companies can charge users fairly based on actual usage while providing predictable and scalable billing.