# Daytona

Sandboxes with no session cap. Disk survives a stop, and GPUs are available.

Route to it: `--provider daytona`.

Compare it with others: [/providers/compare?p=daytona](/providers/compare?p=daytona).

## Price

| Size | Asked for | Per hour | Per month | Runs as | Plan |
| --- | --- | --- | --- | --- | --- |
| S | 1 vCPU · 2 GiB | $0.100/hr | $73/mo | 1 vCPU · 3 GiB · 10 GiB | $0 · $200 credit |
| M | 2 vCPU · 4 GiB | $0.199/hr | $145/mo | 2 vCPU · 6 GiB · 10 GiB | $0 · $200 credit |
| L | 4 vCPU · 8 GiB | $0.397/hr | $289/mo | 4 vCPU · 12 GiB · 10 GiB | $0 + raised limits |
| XL | 8 vCPU · 16 GiB | Not offered | Not offered | — | — |

A month is 730 hours of running. Daytona bills CPU the whole time a sandbox runs, so it is always 100%.

## Limits and features

What works through Sandbox Router:

| Feature | Value | Details |
| --- | --- | --- |
| Isolation | Container + microVM |  |
| Snapshots | Filesystem | memory on VMs |
| Kept after stop | Disk |  |
| When idle | Stops when idle |  |
| Max session | No limit | stops when idle |
| Max vCPU | 4 | raised limits, estimated |
| Max memory | 12 GiB | raised limits, estimated |
| Max disk | 20 GiB | raised limits, estimated |
| Sizing | XS–2XL, or custom |  |
| Custom image | Yes |  |
| GPU | Yes | Direct calls: Yes. The provider has this, but Sandbox Router doesn't pass it through yet. |
| Self-hosting | BYOC | custom regions |

- † On your own key, the provider's limits for your plan apply. Daytona's defaults fit up to M: L needs Daytona to raise your limits. XL (8 vCPU · 16 GiB) is over every provider's limits through Sandbox Router.
- § Daytona applies a host list only in a Tier 3 or higher Daytona organization.
- Shape not confirmed: Sandbox Router may boot Daytona's default sandbox, 1 vCPU · 1 GiB · 3 GiB at every size, so its prices and limits here are estimates.

Pricing sources, checked 6 Nov 2026: [daytona.io/pricing](https://www.daytona.io/pricing), [Billing docs](https://www.daytona.io/docs/en/billing/), [Sandbox limits](https://www.daytona.io/docs/en/sandboxes/).

## Benchmarks

From [ComputeSDK](https://www.computesdk.com/benchmarks/sandboxes), which runs the same test against every provider. Latest run Nov 6, 2026.

- Burst TTI score: 93.2. Rank 9 of 30 on the latest run, up 0.3 from Oct 30. ([source](https://www.computesdk.com/benchmarks/sandboxes/burst-tti/daytona))
- Build time: 81.5 s. Rank 18 of 35 on the latest run, 7.1 s slower than Oct 23. Failed Oct 30. ([source](https://www.computesdk.com/benchmarks/sandboxes/dax/daytona))

| Run | Burst TTI score | Build time |
| --- | --- | --- |
| Nov 6, 2026 | 93.2 | 81.5 s |
| Oct 30, 2026 | 92.8 | Failed |
| Oct 23, 2026 | 87.4 | 74.4 s |
| Oct 20, 2026 | 92.2 | 75.3 s |
| Oct 16, 2026 | 91.9 | 73.1 s |
| Oct 9, 2026 | 92.7 | 70.0 s |
| Oct 2, 2026 | 96.3 | 69.7 s |
| Sep 25, 2026 | 95.9 | 69.2 s |