Pricing

​

Overview

The pricing calculator on the Goldsky website is a good resource for estimating run rate costs.

​

Tiers

Goldsky offers three plans:

• Starter (free) - access to subgraphs + Mirror/Turbo pipelines
• Scale (pay-as-you-go) - access to the above + hosted databases
• Enterprise - access to the above + dedicated support, advanced features (eg. static IP addresses, custom network integrations) and committed-use discounts

All users who sign up and do not enter a credit card are on the Starter plan. Users in this tier will not be able to exceed the free plan limits outlined below. By entering a credit card you automatically upgrade to the Scale plan and are billed on your usage, charged at the end of every month. Enterprise plan customers are charged in accordance with their contracts.

​

Hourly metering

Our prices are quoted on a monthly basis for simpler presentation, but metered and billed on an hourly basis. This has a few key implications:

1. To account for the varying number of days in each month of the year, we conservatively estimate that each month has 730 hours. For hosted databases, the assumption is 720 hours per month.
2. All estimations on this page assume “always-on” capacity. In practice, you can run double the number of subgraph workers or pipeline workers for half the time and pay the same price. This similarly holds for the “entities stored” metric in subgraphs, etc.

​

Billing

Usage across all products is tracked and billed monthly through the Goldsky Dashboard. You can monitor your usage and costs in real-time across all meters including subgraph workers, pipeline workers, storage, bandwidth, and RPC requests. For high-volume pricing or enterprise agreements, schedule a call with our team.

​

Subgraphs

We bill for usage based on two metrics:

1. the number of active subgraphs
2. the amount of data stored across all subgraphs in your project.

​

Compute

The number of active subgraph workers, tracked hourly. If you pause or delete a subgraph, it is no longer billed. One subgraph run for an entire month therefore costs the same as two subgraphs run for half a month.

​

Storage

The number of entities stored across all subgraphs in your project, tracked hourly. If you delete a subgraph, stored entities are no longer tracked. All entities in a project count toward the project’s usage on a cumulative basis.

​

Mirror/Turbo

We bill for usage based on two metrics:

1. the number of active Mirror/Turbo pipeline workers
2. the amount of data written to your sinks by Mirror/Turbo pipelines

​

Compute

The number of active workers, billed hourly. Pipeline resources can have multiple parallel workers, and each worker incurs usage separately. If you have one small pipeline and one large pipeline each deployed for 2 hours, you will accumulate 1*2*1 + 1*2*10 = 2 + 20 = 22 hours of usage. Note: Pipelines that use a single subgraph as a source, and webhooks or GraphQL APIs as sink(s), are not metered as pipelines. However, you still accumulate hourly subgraph usage. Examples:

1. If you have 1 small pipeline, you use 1 pipeline worker-hour every hour. At 730 hours in the average month, you would incur 730 pipeline worker-hours for that month.
2. If you start with 10 small pipelines in a billing period and delete all of them halfway through the billing period, you are charged the equivalent of 5 pipeline workers for the full billing period.
3. If you have 2 large pipelines, you will be using 20 pipeline worker-hours every hour, equating to 14,600 pipeline worker-hours if you run them the entire month.

​

Bandwidth

The number of records written by pipelines in your project. For example, for a PostgreSQL sink, every row created, updated, or deleted, counts as a ‘write’. For a Kafka sink, every message counts as write. Examples:

1. If you have a pipeline that writes 20,000 records per day for 10 days, and then 20 records per day for 10 days, you will be using 200,200 pipeline event writes.
2. If you have two pipelines that each write 1 million events in one month, then you are not charged for the first one million events, but you are charged $1 for the next one million, as per the Starter Plan limits below.

​

RPC Edge

RPC Edge provides high-performance RPC endpoints for EVM networks. Pricing is $5 per million requests with all RPC methods priced equally. Tiered volume discounts are available for usage over 500M requests/month. Learn more about RPC Edge in the Edge documentation.

​

Compose

While in Beta, pricing for Compose is enterprise only. Schedule a call with our team to discuss your needs.

​

Hosted databases

Hosted databases are used for both Mirror as a sink option, and for Compose applications. We bill for usage based on two metrics:

1. the amount of total storage used
2. the amount of compute time (memory and CPU) used

​

Compute

The total amount of active CPU hours used by all databases * number of vCPUs used. This is tracked hourly. If you delete or pause a pipeline that uses a hosted Postgres database, the database will transition to idle mode and you won’t incur utilization charges during that time. Note that if you query the database from an external source, like a DB visualization tool, you will be charged for utilization since the database is actively being queried. VCUs are auto-scaled, so you’ll be charged a variable hourly rate depending on how much time is spent in each VCU range.

​

Storage

We bill for database storage based on the amount of storage you use during your billing cycle, measured hourly.

​

AWS Marketplace

Goldsky is available on AWS Marketplace as an approved AWS Partner (APN). This provides an alternative purchasing option for customers who want to consolidate their cloud spending.

​

Benefits

• Consolidated billing: Goldsky charges appear directly on your AWS invoice
• Use existing AWS commit: AWS Marketplace purchases can count toward retiring up to 25% of your annual AWS commitment
• Simplified procurement: Purchase through your existing AWS account without additional vendor onboarding

To purchase Goldsky through AWS Marketplace, contact our team.