Similar to BigQuery, Substreams is billed off how many bytes the streaming engine (Firehose) has to process over a segment of history or in real-time.

This means, opening up the stored locally Full Block flat files, applying the filters or transactions to the blocks so that they can then be streamed out. This means that you are not billed on egress, the exception to this rule is for Solana Account Changes which are billed solely on your egress usage as opposed to processed bytes.

You don’t need to know how many bytes you’ll need in advance. You will only be charged for what you use.

To keep one stream in real-time depends very much on the chain of interest. Usage may vary per month from 2 TiB up to 30+ TiB for chains like Solana.

If you plan on reprocessing large historical segments and would like to see up to 10x performance boosts then fill out the get in touch or reach out directly over discord.

Billing for historical data can vary based on the stages involved, such as when `store` modules are implemented and read from, or when modules are used as inputs for others. To check the required stages, run: `substreams info your-package.spkg your-module-name`.

Billing is always honoured based on the product you choose at The Graph. Therefore, if you are building a Substreams-powered Subgraph you should only be concerned with the query based billing model in production.

Note if you’re initially running tests on the Substreams component and not making Subgraph queries over the network then you will be billed for the associated Substreams usage.