Fly Machines vs Blaxel: pricing, idle costs and limits

A Fly Machine bills a fixed size while started; Blaxel bills memory with CPU included, and costs nothing for compute in standby.

Runtime runs the same job for 55% less than a Fly Machine and 77% less than Blaxel. 1,000 one-minute runs of a 2 vCPU, 4 GiB sandbox cost $1.44 on a Machine, $2.76 on Blaxel and $0.64 on Runtime. Neither rival's price moves with how busy the CPU is; Runtime bills the CPU your code uses, at $0.025 a vCPU-hour. Rates checked 23 September 2026.

At a glance

The Machine is a performance-2x with 4 GB, at the rates of Fly's Ashburn (iad) region.

	Fly Machines	Blaxel	Runtime
Isolation	Firecracker microVM	Lightweight virtual machines	Firecracker microVM, own kernel
CPU	By size while started: $0.0861 an hour for 2 vCPUs and 4 GB	Included with memory; 8 GB gets 4 cores	$0.025 per vCPU-hour of measured CPU, with a small floor
Memory	Included in the size; more at $5 per GB a month	$0.0000115 per GB-second of active time	$0.0075 per reserved GiB-hour
Idle	Billed until stopped or suspended	Standby about 15 s after connections close	Pause on call or after idlePauseSeconds
Kept idle	Files on volumes; memory with suspend, at 2 GB or less	Files and running processes	Files, memory and running processes, 1 to 365 days
Resume	A few hundred ms from suspend	Under 25 ms from standby	Usually about half a second
Idle storage	Storage only; volumes billed separately	$0.20 per GB-month of snapshot	$0.08 per GB-month
Plan fee	None; pay as you go	None; tiers grow with credit added	None; prepaid credit from $10
Free start	Not stated on the pricing page	Up to $200 of credit	50 sandbox hours, no card
Interfaces	Machines API and flyctl	TypeScript and Python SDKs	API, CLI, MCP server, JavaScript and Python SDKs

Which one costs less when the agent sits idle?

Blaxel, because it stops charging by itself. A Blaxel sandbox goes to standby about 15 seconds after its last connection closes, with no memory charge there, only the snapshot's storage. Reconnecting brings it back in under 25 ms with its files and running processes. Blaxel notes that external connections, such as database connections, message queues and HTTP pools, are not kept.

A Fly Machine bills its whole size for every second it is started. To stop paying, stop it, and a stopped Machine costs storage only; but a stop resets its root filesystem, so files that must last go on a Fly Volume. Suspend keeps memory and resumes in a few hundred ms, at the same storage-only cost, but Fly lists 2 GB of memory or less as a requirement, so the 4 GB Machine priced here cannot use it.

While it runs, the Machine is the cheaper of the two: $1.44 against $2.76 for the job below. The answer turns on how long each sandbox sits idle before you stop it, and whether memory has to survive.

Cost for the same job

1,000 runs of a 2 vCPU, 4 GiB sandbox (4 GB on Blaxel and the Machine). Each run lasts 60 seconds and keeps the CPU busy for 20 CPU-seconds: an agent that mostly waits for a model.

TextMachines  Size   1,000 × 60 s / 3,600 × $0.0861      = $1.44Blaxel    Memory 1,000 × 60 s × 4 × $0.0000115       = $2.76Runtime   CPU    1,000 × 20 s / 3,600 × $0.025       = $0.14          Memory 1,000 × 60 s / 3,600 × 4 × $0.0075  = $0.50          Total                                         $0.64

At 100,000 runs a month that is $143.50 on Machines, $276.00 on Blaxel and $63.89 on Runtime. With both CPUs busy for the whole minute, it is $1.33 on Runtime against $1.44 on Machines and $2.76 on Blaxel, so Runtime is cheaper however busy the sandbox is. The Machine figure assumes each Machine is stopped when its run ends, and the Blaxel figure that each sandbox goes to standby the moment its run ends; Blaxel's 15 seconds would add $0.69 per 1,000 runs. Storage, network, plan fees, taxes and free credit are left out.

Pause and wake on Runtime

A Runtime sandbox pauses when you call pause(), or by itself after idle_pause_seconds with no request. Compute billing stops, memory and files are kept, and the sandbox wakes on the next request, or when you call wake(), from any process:

Pythonfrom withruntime import Sandboxsbx = Sandbox.create(timeout_seconds=600)sbx.exec("echo state > /workspace/state.txt")sbx.pause()  # memory and files are kept; compute billing stopsagain = Sandbox.connect(sbx.id)again.wake(timeout_seconds=1200)again.stop()

See how to pause and resume a sandbox and what a cold start is.

Which one to pick

• Pick Fly Machines for general-purpose VMs in 18 regions, up to 16 performance vCPUs and 128 GB, with volumes and Managed Postgres in the same account, or for CPU-bound jobs at a flat hourly price.
• Pick Blaxel for sandboxes that idle for free and come back in about 25 ms, or very large fleets: its top tier runs over 100,000 sandboxes at once.
• Pick Runtime for the lowest bill of the three at any load, CPU and memory chosen apart, and paused sandboxes kept with their memory at $0.08 per GB-month.

Switch from either one

From Machines, if you create a Machine for each job through the Machines API, create a Runtime sandbox instead and stop it when the job ends. From Blaxel, standby maps to Runtime's automatic wake, which is on by default: a request to a paused sandbox wakes it.

Or give your coding agent the one instruction in migration: it switches the project on a branch, tests it on the free trial and reports the monthly saving. From Fly, npx withruntime switch --from fly before your first top-up matches that top-up with credit, up to $100. The full side-by-sides are Runtime vs Fly.io and Runtime vs Blaxel; Blaxel against Cloudflare is in Cloudflare Sandbox vs Blaxel.

Sources

Rates checked 23 September 2026; Fly's suspend page and Blaxel's sandbox overview read 25 September 2026.

• Fly.io pricing, Machine suspend and resume, Fly Volumes, Fly.io regions
• Blaxel pricing, Blaxel sandboxes
• Runtime pricing

Facts on this page were checked on 25 September 2026.