Blaxel vs Northflank: pricing, isolation and limits

Northflank bills allocated CPU and memory at low hourly rates; Blaxel bills memory only while active. The same job is $1.11 against $2.76.

Runtime runs the same job for 77% less than Blaxel and 42% less than Northflank. 1,000 one-minute runs of a 2 vCPU, 4 GiB sandbox cost $2.76 on Blaxel, $1.11 on Northflank and $0.64 on Runtime, because Runtime bills the CPU your code uses rather than the CPUs it holds. Rates checked 23 September 2026.

At a glance

	Blaxel	Northflank	Runtime
Isolation	Lightweight virtual machines	Kata Containers and gVisor	Firecracker microVM, own kernel, every sandbox
CPU	Included with memory: 8 GB gets 4 cores	$0.01667 per allocated vCPU-hour	$0.025 per vCPU-hour of measured CPU, with a small floor
Memory	$0.0000115 per GB-second of active time	$0.00833 per GB-hour	$0.0075 per reserved GiB-hour
2 vCPU, 4 GB, an hour	$0.1656 while active	$0.0667 while running	$0.03125 waiting, $0.08 fully busy
Idle	Standby after about 15 s; no compute charge	Runs until you stop it; no forced time limit	Pause keeps files, memory and processes, 1 to 365 days
What you create	A sandbox	A service with a deployment plan	A sandbox, in one call
Plan fee	None; tiers grow with credit added	None published for compute	None; prepaid credit from $10
Free start	Up to $200 of credit	2 free services and 1 free database	50 sandbox hours, no card
Where it runs	Blaxel's own regions	Northflank's regions, or your AWS, GCP or Azure account	Dedicated servers Runtime operates

Standby or always on: which is cheaper?

It depends on how much of the day the sandbox works. A 2 vCPU, 4 GB service on Northflank costs $0.0667 an hour, billed for every hour it runs. Northflank presents running for weeks as a feature ("no forced time limits"); its free Developer Sandbox is "always-on compute, no sleeping".

Blaxel charges $0.1656 an hour for 4 GB while the sandbox is active, about two and a half times as much, but nothing for compute once it is on standby, which starts about 15 seconds after the last connection closes. Standby costs snapshot storage, $0.20 per GB-month. On compute alone, Blaxel is the cheaper of the two when a sandbox is active for less than about 40% of the hours Northflank would keep it running.

Runtime does both. A waiting sandbox is billed only the CPU floor, a twentieth of a vCPU, so 2 vCPU and 4 GiB costs $0.03125 an hour while it waits on a model, under half of Northflank's figure. Set idlePauseSeconds and it pauses itself, keeping memory and processes for $0.08 per GB-month, and the next request wakes it.

Cost for the same job

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

TextBlaxel      Memory 1,000 × 60 s × 4 × $0.0000115       = $2.76Northflank  CPU    1,000 × 60 s / 3,600 × 2 × $0.01667 = $0.556            Memory 1,000 × 60 s / 3,600 × 4 × $0.00833 = $0.555            Total                                         $1.11Runtime     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 $276.00 on Blaxel, $111.10 on Northflank and $63.89 on Runtime. With both CPUs busy for the whole minute, it is $1.11 on Northflank, below $1.33 on Runtime, and $2.76 on Blaxel. Northflank's allocated rate wins when every CPU works all the time. The Blaxel figure assumes standby the moment each run ends. Storage, network, taxes and free allowances are left out.

How do they isolate code?

Northflank states that its sandboxes use Kata Containers and gVisor for VM-level isolation. Blaxel describes its sandboxes as lightweight virtual machines and does not name the hypervisor. Runtime gives every sandbox a Firecracker microVM with its own Linux kernel. See Kata Containers vs Firecracker and Firecracker vs gVisor for how those boundaries differ.

Which one to pick

• Pick Blaxel if your sandboxes are idle most of the time and must come back fast: Blaxel states a resume in under 25 ms, and its top tier runs over 100,000 sandboxes at once.
• Pick Northflank for sandboxes that run for days with every CPU busy, for GPUs (L4, A100 and H100 by the hour), or to run in your own AWS, GCP or Azure account.
• Pick Runtime for agents that wait on a model: measured CPU makes the typical run 42% cheaper than Northflank and 77% cheaper than Blaxel, and a sandbox is one call, with pause and forks that keep memory.

Switch from either one

Neither has a drop-in import on Runtime; the calls map directly (map the calls). Blaxel's SandboxInstance.create() becomes Sandbox.create() and its delete() becomes stop(). A Northflank sandbox service with its deployment plan becomes one create call with the size you want:

TypeScriptimport { Sandbox } from "withruntime";const box = await Sandbox.create({ funding: "trial" });try {  await box.files.write("/workspace/task.py", "print(6 * 7)\n");  console.log((await box.exec("python3 /workspace/task.py", { check: true })).stdout);} finally {  await box.stop();}

Pythonfrom withruntime import Sandboxwith Sandbox.create(funding="trial") as box:    box.files.write("/workspace/task.py", "print(6 * 7)\n")    print(box.exec("python3 /workspace/task.py", check=True).stdout)

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. npx withruntime compare --from northflank (or --from blaxel) prices what you ran at the old provider's rates. The full side-by-sides are Runtime vs Blaxel and Runtime vs Northflank.

Sources

Rates checked 23 September 2026. Both pricing pages, Blaxel's sandbox documentation and Northflank's sandbox page reread 25 September 2026.

• Blaxel pricing, Blaxel sandboxes
• Northflank pricing, Northflank sandboxes, Sandboxes on Northflank
• Runtime pricing and security

Facts on this page were checked on 25 September 2026.