Fly.io
Application server software
- Features
- Ease of use
- Ease of management
- Quality of support
- Affordability
- Market presence
Take the quiz to check if Fly.io and its alternatives fit your requirements.
Pay-as-you-go
Free Trial
Free version unavailable
Small
Medium
Large
- Accommodation and food services
- Arts, entertainment, and recreation
- Education and training
What is Fly.io
Fly.io is a platform for running application services on a global edge infrastructure using containers and lightweight virtual machines. It targets developers and teams that want to deploy web apps, APIs, and background workers close to users with built-in networking and operational tooling. The service emphasizes region-aware deployment, private networking between services, and managed primitives such as volumes for stateful workloads. Fly.io is typically used as an alternative to managing application servers and multi-region networking directly on IaaS.
Global, region-aware deployments
Fly.io supports deploying the same application to multiple regions and routing traffic based on proximity and health. This can reduce latency for globally distributed users without requiring teams to build their own multi-region load balancing. The platform provides primitives for controlling regions, scaling, and process groups so teams can separate web and worker roles. This is a differentiator versus traditional single-datacenter application server setups.
Container-based runtime model
Applications deploy as container images, which fits common CI/CD workflows and language stacks. The platform runs workloads on lightweight VMs, which can simplify isolation compared with shared-process hosting models. Teams can bring their own Dockerfiles and runtime dependencies rather than conforming to a single application server framework. This makes it suitable for heterogeneous microservices and polyglot environments.
Integrated networking and ops tooling
Fly.io includes built-in service discovery, private networking between apps, and TLS/edge routing features that reduce the need for separate networking components. It provides a CLI and configuration model for repeatable deployments and environment management. Operational features such as health checks, scaling controls, and logs/metrics integrations support day-to-day running of services. This can reduce the amount of separate infrastructure software teams must assemble.
Platform-specific operational model
Fly.io uses its own concepts (apps, regions, process groups, Machines, volumes) that teams must learn and operationalize. Migrating workloads to or from the platform may require changes to deployment pipelines and networking assumptions. Organizations that prefer standardized server management patterns may find the model less familiar than conventional application server stacks. This can increase onboarding time for operations and security teams.
Stateful workloads need planning
While Fly.io supports persistent volumes, stateful services require careful design around replication, failover, and region placement. Multi-region state introduces trade-offs in consistency and operational complexity that the platform does not fully eliminate. Teams may still need external managed databases or additional tooling for high availability. This is a common constraint when running stateful components near the edge.
Less control than self-managed servers
As a managed platform, Fly.io abstracts underlying infrastructure choices and limits certain low-level tuning compared with running your own application servers on IaaS. Some advanced networking, compliance, or custom observability requirements may be harder to satisfy without platform support. Troubleshooting can depend on provider tooling and support processes rather than direct host access. This may be a consideration for regulated environments or highly customized stacks.
Plan & Pricing
Pricing model: Pay-as-you-go Free tier/trial: Free trial available (see notes); no permanent free plan for new organizations (legacy free allowances remain only for older/deprecated plans).
Compute (example presets, started Machines)
- shared-cpu-1x (1 shared CPU)
- 256 MB: $0.00000075/sec | $0.0027/hr | $1.94/month
- 512 MB: $0.00000123/sec | $0.0044/hr | $3.19/month
- 1 GB: $0.00000220/sec | $0.0079/hr | $5.70/month
- 2 GB: $0.00000413/sec | $0.0149/hr | $10.70/month
- shared-cpu-2x (2 shared CPU)
- 512 MB: $0.00000150/sec | $0.0054/hr | $3.89/month
- 1 GB: $0.00000246/sec | $0.0089/hr | $6.39/month
- 2 GB: $0.00000440/sec | $0.0158/hr | $11.39/month
- 4 GB: $0.00000826/sec | $0.0297/hr | $21.40/month
- Additional RAM: ~ $5 per 30 days per additional GB (baseline; region markups may apply).
Stopped Machines
- Root filesystem (rootfs) for stopped Machines: $0.15 per GB per 30 days.
Storage & snapshots
- Volumes (persistent storage): $0.15/GB per month (pro-rated to the hour).
- Volume snapshots: $0.08/GB per month; first 10 GB free each month.
Networking & certificates
- Dedicated IPv4 addresses: $2/month.
- Managed SSL certificates: Single hostname $0.10/month; Wildcard $1/month. (Each organization's first 10 single-hostname certs are free.)
- Data transfer: Outbound data (egress) billed per region group; inbound data transfer is free. (Regional rates apply.)
Managed services & extras
- Fly Kubernetes (FKS): $75/month per cluster + cost of compute and volumes used.
- Extensions: Third-party managed services billed at their list prices via Fly billing (charges shown in dashboard).
Support & compliance
- Support plans (add-on): Standard $29/month; Premium $199/month; Enterprise starting at $2,500/month.
- Compliance (HIPAA/BAA/SOC2 support): $99/month (add-on).
Machine reservation (prepay) discounts
- 40% discount via reservation blocks (examples):
- Performance: $144/year (gives $20/mo credit), $1,440/year (gives $200/mo credit), $14,400/year (gives $2,000/mo credit).
- Shared: $36/year (gives $5/mo credit), $360/year (gives $50/mo credit), $3,600/year (gives $500/mo credit).
GPU machines
- GPUs are deprecated (will be unavailable after the stated deprecation date) but on-demand GPU pricing (when available) was published as examples (A10 $0.75/hr, L40S $0.70/hr, A100 variants $1.25–$1.50/hr); GPUs are billed separately when attached.
Free trial (explicit)
- New users can use a free trial: 2 total VM hours OR 7 calendar days (whichever comes first). Trial limits: max 10 machines, 20 GB volume storage, up to 2 vCPUs per machine, 4 GB memory per machine. Adding a credit card ends the free trial and billing begins.
Notes & caveats
- Pricing is per-organization and pro-rated. Region-specific markups apply; use Fly.io pricing calculator or dashboard for an exact estimate. Some legacy “free allowances” and deprecated plans remain for older organizations but are not offered to new organizations.
Seller details
Fly.io, Inc.
Chicago, IL, USA
2017
Private
https://fly.io
https://x.com/flydotio
https://www.linkedin.com/company/flydotio/
