WunderGraph Cosmo

Overview

Cosmo brings GraphQL, REST, gRPC and Kafka into one federated graph. It supports Apollo Federation 1 and 2 and works with existing GraphQL services across frameworks and languages, which makes it relevant to teams extending an established service landscape rather than adopting a single framework.

The Go-based Cosmo Router is Apache 2.0-licensed and can run in Docker, Kubernetes or an AWS Lambda-optimized binary, or through WunderGraph’s managed service. That deployment choice is a real strength: teams can retain operational control or delegate router hosting. The trade-off is that the generous governance and operational limits are tied to paid plans.

Key features

Schema governance and observability

The Schema Registry checks composition, breaking changes, client traffic and linting. These guardrails help teams catch graph changes that could disrupt consumers before they propagate. OpenTelemetry tracing and Prometheus metrics can be exported to compatible monitoring backends, giving teams ways to connect Cosmo’s telemetry to their monitoring setup.

Subscriptions and integrations

Subscriptions work over WebSockets and Server-Sent Events, with event-driven subscriptions backed by Kafka, NATS, SQS or RabbitMQ. Cosmo Streams currently supports Kafka, NATS, including JetStream, and Redis. The range is useful for teams combining request-response APIs with event-driven services, though the Streams integrations cover a narrower set of systems than the subscription backends.

Privacy and compliance

Router defaults are designed to keep sensitive request information out of the control plane and analytics. Managed plans include SOC 2 and HIPAA compliance, and WunderGraph identifies ISO 27001 certification. Teams should distinguish those managed-plan assurances from the self-hosted deployment option.

Pricing

Cosmo is freemium, with no free trial. The free Developer plan is billed Free forever and includes one federated graph, 10M requests per month, 10 subgraphs, one team member, seven-day retention and community support. That is a useful way for an individual to start a graph, but a single seat and short retention period sharply limit its fit for a collaborative team.

Launch costs $499 / month and includes two federated graphs, 250M requests per month, 20 subgraphs, 10 team members, 14-day retention and email support. It is the practical entry point for a small team that needs shared access and materially higher quotas, though retention remains brief.

Scale costs $3499 / month for five graphs, 1B requests per month, 100 subgraphs, 25 team members and 30-day retention. It adds SSO/RBAC and a 99% managed-service SLA, making it better suited to larger teams that need access controls and a defined service commitment. Enterprise has custom pricing, custom graph, request, retention and subgraph limits, unlimited team size, SCIM, a dedicated solution architect and an SLA of up to 99.99%. Its terms suit organizations that need tailored capacity or more support, but the price requires a custom arrangement.

Platforms

Cosmo is available for API, self-hosted and web use. The router’s Docker, Kubernetes and AWS Lambda-optimized deployment options give infrastructure teams several ways to operate it themselves, while the managed service offers an alternative. Server-language support spans Go, Python, Java, C#, Node.js and Rust.

Who it's for

Cosmo is aimed at teams scaling from their first subgraph to thousands of daily operations. It is strongest for platform teams that want federation alongside schema governance and observability, especially when they need deployment flexibility or a single graph spanning several protocols. It is less suitable for an individual or small group unwilling to accept the free plan’s one-seat, seven-day-retention ceiling or the cost of a paid tier.

Pros and cons

• Pros: Apollo Federation 1 and 2 compatibility lets existing GraphQL services fit into the graph across languages and frameworks.
• Pros: Schema checks, tracing and metrics put governance and operations alongside routing rather than treating federation as routing alone.
• Pros: Self-hosted and managed deployment options let teams choose how much infrastructure responsibility to retain.
• Cons: Developer allows only one team member and seven-day retention, so collaboration and historical visibility quickly require an upgrade.
• Cons: The first paid tier is $499 / month, a substantial step for teams that have outgrown the free caps but do not need large quotas.
• Cons: The 99% managed-service SLA appears at Scale, while Enterprise is needed for an SLA up to 99.99% and custom capacity.

Alternatives

For a broader directory of options, see GraphQL Tools.

• GraphQL Hive is worth considering if a free forever plan with unlimited seats, projects, organizations and schema pushes matters more than Cosmo’s stated federation and routing scope; its Hobby plan caps usage at 1M operations per month and retains usage data for seven days.
• Hot Chocolate is a free, MIT-licensed option for teams seeking a GraphQL tool they can use in commercial or private projects.
• Strawberry GraphQL is a free, MIT-licensed open-source Python library for teams looking for a Python GraphQL library rather than Cosmo’s multi-protocol federation platform.
• GraphQL Ruby is a free Ruby gem implementing the GraphQL server specification; choose it when a Ruby server implementation is the priority.
• Altair GraphQL Client is an alternative for people seeking a GraphQL client across desktop, web and extension platforms rather than a federation platform.
• Apollo Client is a free option for client-side GraphQL work across web, Android and iOS, rather than Cosmo’s federation routing and governance.
• Relay is a free alternative for GraphQL use across web and mobile platforms.
• GraphQL Kotlin is a free, Apache License 2.0 option for teams seeking a self-hosted or API-focused GraphQL tool.

Verdict

Choose WunderGraph Cosmo if your platform team needs federation routing, schema guardrails and observability in one system, with the freedom to self-host or use a managed service. Its main advantage is the breadth of operational capability around a federated graph; its main drawback is the sharp gap between the constrained free plan and a $499 / month starting paid tier. Teams that need collaborative use or longer retention should budget for paid service, while those seeking a free GraphQL library or client should look elsewhere.