# Apollo GraphOS vs WunderGraph Cosmo

Canonical: https://inetgeek.com/compare/apollo-graphos-vs-wundergraph-cosmo/

Every value below is read from Apollo GraphOS's and WunderGraph Cosmo's own documentation. See https://inetgeek.com/methodology/ for how.

## At a glance

| Criterion | Apollo GraphOS | WunderGraph Cosmo |
| --- | --- | --- |
| Free tier | Yes | Yes |
| What the free plan includes | A self-hosted router rate-limited to 60 requests per minute, up to 3 developers, one day of retention for checks and insights, and community support. Federation, connectors, response caching and federated subscriptions are all included at this tier. | One federated graph and up to 10 million requests a month, with the self-hosted router and Discord community support. Free forever rather than a trial. |
| Entry paid plan | Developer starts at $5 per million requests, billed monthly, with a $0 plan base price and automatic volume discounts — a per-request rate rather than a plan fee, so no magnitude is recorded. New signups get $50 of usage credit and a card is required. | $499/month for Launch: two federated graphs, up to 250 million requests a month and email support. Scale is $3,499/month for five graphs and up to 1 billion requests. |
| Pricing model | Per request, with a $0 plan base. Every query and mutation counts as a standard operation whichever runtime processes it, and subscription update events are metered independently on top. | Flat monthly plans bounded by two things at once: how many federated graphs you run, and how many requests a month they serve. Both ceilings rise together tier by tier. |
| Self-hostable | Supported | Supported |

## Where they differ

### What the free plan includes

- Apollo GraphOS: A self-hosted router rate-limited to 60 requests per minute, up to 3 developers, one day of retention for checks and insights, and community support. Federation, connectors, response caching and federated subscriptions are all included at this tier. ([source](https://www.apollographql.com/pricing))
- WunderGraph Cosmo: One federated graph and up to 10 million requests a month, with the self-hosted router and Discord community support. Free forever rather than a trial. ([source](https://wundergraph.com/pricing))

### Entry paid plan

- Apollo GraphOS: Developer starts at $5 per million requests, billed monthly, with a $0 plan base price and automatic volume discounts — a per-request rate rather than a plan fee, so no magnitude is recorded. New signups get $50 of usage credit and a card is required. ([source](https://www.apollographql.com/pricing))
- WunderGraph Cosmo: $499/month for Launch: two federated graphs, up to 250 million requests a month and email support. Scale is $3,499/month for five graphs and up to 1 billion requests. ([source](https://wundergraph.com/pricing))

### Pricing model

- Apollo GraphOS: Per request, with a $0 plan base. Every query and mutation counts as a standard operation whichever runtime processes it, and subscription update events are metered independently on top. ([source](https://www.apollographql.com/pricing))
- WunderGraph Cosmo: Flat monthly plans bounded by two things at once: how many federated graphs you run, and how many requests a month they serve. Both ceilings rise together tier by tier. ([source](https://wundergraph.com/pricing))

## Which should you choose?

Pick Apollo GraphOS if A federated graph whose traffic is spiky or unknown, where a $0 plan base and $5 per million requests means the bill tracks usage instead of a tier — and where metering subscription update events separately is a feature rather than a surprise.

Pick WunderGraph Cosmo if Teams who would rather buy a ceiling than a meter: $499 a month covers two federated graphs and 250 million requests, with the router self-hosted and nothing counted per query.

Consider something else: Apollo GraphOS — You want a fixed number to budget. Apollo's price moves with every query and mutation whichever runtime processes them, and subscription events are counted on top; a flat plan at a higher headline figure can cost less and is easier to forecast.

Consider something else: WunderGraph Cosmo — You run one small graph, or you need SSO and RBAC. The free tier stops at 10 million requests a month and one graph, and both SSO and RBAC first appear on Scale at $3,499 a month — seven times the entry price.

## Documented by only one

- Real-time subscriptions: Apollo GraphOS: Federated subscriptions on every tier including the free one, but metered separately: update events are counted independently of standard operations and priced per million.
- Support on entry paid plan: WunderGraph Cosmo: Discord community support on the free Developer tier; email support from Launch ($499/month); Slack Connect from Scale ($3,499/month), which is also where SSO, RBAC and a standard SLA appear.
- Role-based access control: WunderGraph Cosmo: SSO and RBAC appear at the Scale tier ($3,499/month), alongside 30 days of data retention and a standard SLA. Neither is listed on Developer or Launch.

## Questions this comparison answers

**Entry paid plan: Apollo GraphOS or WunderGraph Cosmo?**

Apollo GraphOS: Developer starts at $5 per million requests, billed monthly, with a $0 plan base price and automatic volume discounts — a per-request rate rather than a plan fee, so no magnitude is recorded. New signups get $50 of usage credit and a card is required.
WunderGraph Cosmo: $499/month for Launch: two federated graphs, up to 250 million requests a month and email support. Scale is $3,499/month for five graphs and up to 1 billion requests.

**Should I choose Apollo GraphOS or WunderGraph Cosmo?**

Pick Apollo GraphOS if A federated graph whose traffic is spiky or unknown, where a $0 plan base and $5 per million requests means the bill tracks usage instead of a tier — and where metering subscription update events separately is a feature rather than a surprise.
Pick WunderGraph Cosmo if Teams who would rather buy a ceiling than a meter: $499 a month covers two federated graphs and 250 million requests, with the router self-hosted and nothing counted per query.