Isorun
$19.36 a month*
Home / AI agent sandboxes / Modal Sandboxes
Containers for untrusted code on Modal's serverless platform, billed per second on max(request, actual usage).
Verdict (2026-11-14): Modal Sandboxes scores 57/100, #47 of 63 AI agent sandboxes. Strong on free tier, docs and SDKs. Best for: Bursty agent workloads where average CPU and memory use is far below peak, since billing follows actual usage when no request is set.
Skip it if: Sandbox CPU and memory cost 3x Modal's Function list rates.
Prices checked 2026-11-15.
Modal Sandboxes are containers created from the Python, JavaScript or Go SDK that run on either a gVisor runtime or a VM runtime with its own Linux kernel. They are billed per second at Modal's non-preemptible CPU and memory rates (3x the Function list price), charged on whichever is higher of the requested or actually used resources, so an unrequested sandbox pays mostly for what it uses. Sandboxes can attach GPUs (gVisor runtime only), run up to 24 hours, and save state through filesystem, directory and memory snapshots. Plans add a monthly compute credit: $30 on Starter and $100 on the $250/month Team plan.
57/100 in AI agent sandboxes, rank 47 of 63. Parts with no data are left out and the rest is rescaled.
Full Modal Sandboxes pricing: plans and costs at four workloads →
$19.36 a month*
$11.73 a month*
$10.14 a month*
$19.00 a month*
$17.95 a month*
$28.44 a month*
$20.00 a month*
$20.00 a month*
| Benchmark | Test | Result | Rank | Data from |
|---|---|---|---|---|
| ComputeSDK benchmarks | Burst start time (median of 100 at once) | 451 ms | 10 of 27 | 2026-11-13 |
| ComputeSDK benchmarks | Repo clone, install and typecheck (one run) | 48,147 ms | 4 of 25 | 2026-11-13 |
| StarSling HPC sandbox benchmarks | Node.js web tooling (runs per second) | 15.54 runs/s | 8 of 12 | 2026-09-29 |
| StarSling HPC sandbox benchmarks | Better Auth cold dependency install | 14.8 s | 5 of 11 | 2026-09-29 |
| Modal Sandboxes | |
|---|---|
| Category | agent sandbox, GPU sandbox |
| Our score | 57/100 |
| Isolation | gVisor container runtime, or a VM runtime with its own Linux kernel (selectable per sandbox) |
| vCPU per hour | $0.07096 |
| Memory per GiB-hour | $0.02401 |
| Storage per GiB-month | $0.09 |
| Billing | per second |
| Free tier | Starter plan: $0/month with $30 of compute credit every month (CPU, memory, GPU for Functions and Sandboxes). |
| Pause and resume | No |
| Snapshots | Yes |
| GPU | Yes |
| Custom images | Yes |
| Network controls | Yes |
| Max session | 24 hours |
| Open source | No |
| Self-host | No |
| SDKs and interfaces | Python, JavaScript/TypeScript, Go |
| Regions | Runs in Modal's multi-cloud fleet; optional region pinning (us, eu, ap and narrower regions such as us-west, uk, jp) at a 1.15x (broad) or 1.75x (narrow) price multiplier. |
| Facts checked | 2026-11-14 |
No reviews yet. Have you used Modal Sandboxes? Your review helps other developers choose.
Modal Sandboxes scores 57/100, #47 of 63 AI agent sandboxes. Strong on free tier, docs and SDKs. Best for: Bursty agent workloads where average CPU and memory use is far below peak, since billing follows actual usage when no request is set. Facts checked 2026-11-14.
$41.88 a month for 2 vCPU, 4 GiB, 176 hours a month (8 h × 22 days), CPU busy 30%. Free tier: Starter plan: $0/month with $30 of compute credit every month (CPU, memory, GPU for Functions and Sandboxes).
gVisor container runtime, or a VM runtime with its own Linux kernel (selectable per sandbox).
In AI agent sandboxes, the top-rated are Isorun (78/100), Northflank (78/100), Koyeb Sandboxes (76/100). Modal Sandboxes is #47 of 63.
Sandboxes are billed at non-preemptible rates of 3x list price.
Physical core (2 vCPU equivalent) $0.00003942
Memory (non-preemptible) $0.00000667
Nvidia H100 SXM5 $0.001097
$250 + compute / month
$30 / month free credits
$100 / month free credits
Volumes $0.09
For CPU and memory, you'll be charged based on whichever is higher: your request or actual usage.
Disk requests are billed by increasing the memory request at a 20:1 ratio.
You can change this by passing a timeout of up to 24 hours to the Sandbox.create(...) function.
By default, each Sandbox requests 0.125 CPU cores and 128 MiB of memory
Unlike CPU Sandboxes, GPU Sandboxes are subject to preemption.
A multiplier on top of our base usage pricing will be applied to any Function or Sandbox that has a container region defined.
Modal provides three levels of outbound network restriction
Memory Snapshots expire 7 days after creation and cannot currently be extended.