Understand your model costs
Input, output, and context are different. Here is how to compare them.
Two rates, one request
Input tokens include the prompt and conversation history you send. Output tokens are generated by the model. A model can have different rates for each, so compare both prices.
Estimate before you build
Multiply input tokens by the input rate per million, then add output tokens multiplied by the output rate per million. The pricing page has a calculator that uses current catalog rates.
Keep context intentional
Sending an entire conversation again increases input usage. Include the context needed for the task and set an appropriate output limit.
What free means
ModelMesh marks a model Free only when both its input and output rates are zero. A model name containing the word free does not override its configured rates.