Layer 2.5 · head to head
Fly.io vs Modal
Fly.io retired its GPUs on 31 Jul 2026; Modal still rents them, billed by the second.
Fly.io retired its GPU Machines on 31 Jul 2026
Fly's pricing page carries no GPU line on 9 Oct 2026, fly.io/docs/gpus/ redirects to the February 2025 post 'We Were Wrong About GPUs', and a Fly email quoted on Fly's community forum (11 Feb 2026) says GPUs are fully deprecated as of 31 Jul 2026.
Checked 9 Oct 2026.
| Fly.io | Modal | |
|---|---|---|
| Can you leave | Movable with effort. Expect to redo images, storage wiring and networking. | Sticky. Leaving means rewriting against a different interface, not changing a hostname. |
| Counterparty | The GPU product was an extension of an established application platform, and Fly has since withdrawn it. | Low financially; high in code. Modal's programming model is genuinely delightful and genuinely non-portable — that is the trade. |
| What it is | Purpose-built GPU cloud. Owns or leases its own datacentre capacity and sells it directly. | You never see a GPU. You send a request and pay per token or per second of execution. |
| How you reach it | You push a container image and it runs. No cluster to operate. | You supply code; scaling and idle time are the provider's problem. |
| Accelerators | None since 31 Jul 2026. Fly.io rented A100 and L40S GPU Machines; its pricing page lists no GPUs on 9 Oct 2026. | H100, A100, L40S, T4, and others |
| Regions | No GPU regions since the retirement. | US, EU |
| Pricing model | No GPU pricing: GPUs are no longer on Fly's pricing page (read 9 Oct 2026). When offered, GPU Machines were billed per second and scaled to zero. | Per second of GPU time, scale to zero, no idle charge. |
| Getting started | Not available: the GPU product has been retired. | Self-serve, generous free tier historically. |
| Capacity | None. GPU Machines were deprecated as of 31 Jul 2026, according to a Fly email quoted on Fly's own community forum on 11 Feb 2026. | Managed pool. |
| Ownership | Private. | Private. |
Fly.io
For: Nobody new: Fly.io no longer rents GPUs. Teams that ran inference on Fly GPUs need another provider.
The catch: The GPU product is gone. fly.io/docs/gpus/ now redirects to Fly's February 2025 post 'We Were Wrong About GPUs'.
Economics: Not applicable since the retirement.
Modal
For: Python teams who want infrastructure to disappear and are happy writing to a framework to get it.
The catch: Your deployment becomes Modal-shaped. Migrating off is a rewrite, not a redeploy. Price that in on day one, not year two.
Economics: Per-second billing with no idle cost is the cheapest possible shape for bursty inference. It is the wrong shape for a job that runs continuously.
Neither table row is a price
Deliberately. Published on-demand rates at this layer move weekly, and essentially nobody signing a real contract pays them — every serious buyer pays less than every list figure either of these companies publishes. Quoting one here would date this page within a month.
The GPU cost calculator carries dated, sourced figures instead, and the durable finding there is the spread: the identical H100 rents from $1.99 to $11.06 an hour on published on-demand prices, depending only on who you rent it from.
The layers underneath both
Whichever you pick is renting you chips in a building that needs power. In 2026 that is the constraint that binds: Microsoft has disclosed an Azure backlog it cannot fill for want of megawatts rather than accelerators, and the US interconnection queue exceeds 2,600 GW with roughly 80% of projects withdrawing before they energise.
Layer 2 — Silicon · Layer 1 — Energy · The interconnection queue