Navigating the world of cloud services can feel like wandering through a maze, especially when it comes to pricing. Fly.io offers a unique approach that focuses on usage rather than rigid plans, making it both flexible and user-friendly for developers and organizations alike.
At its core, Fly.io bills per organization. This means that if you’re part of a larger team or company, your resource usage is reported back to the parent Billing Organization. It’s designed this way so you can pick and choose what pieces are essential for your application without being bogged down by unnecessary costs.
One key aspect to understand is how billing works based on resource provisioning. Organizations need to have a credit card on file (except Linked Organizations), which allows them to manage their apps effectively while keeping track of expenses as they scale up or down in response to demand.
When we talk about machines—Fly's virtual servers—the pricing varies significantly depending on whether they're started or stopped. For instance, running a shared-cpu-1x1 machine with 256MB RAM will cost just $0.00000075 per second! If you're looking at more robust options like performance-16x16 with 128GB RAM, expect around $0.37664 per second.
Here’s where things get interesting: additional RAM incurs an extra charge—approximately $5 every 30 days for each GB beyond what's included in your preset configuration. This modularity allows users not only flexibility but also control over their expenditures based on actual needs rather than estimates.
If you're planning ahead and know you'll be using certain resources consistently, consider reserving compute time—a smart move that could save you up to 40%. Reservations allow users access to discounted rates across various CPU classes in specific regions; these credits apply monthly but don’t roll over if unused.
For those who require GPU capabilities—think AI workloads or heavy computational tasks—the costs increase accordingly since GPUs are billed separately from standard machines at rates ranging from $1.25/hr for L40S models up to $3.50/hr for high-end A100s.
In essence, understanding Fly.io's pricing structure revolves around recognizing how much power you truly need versus what you're willing—or able—to spend month-to-month.