# StateSet Sandbox

Official pricing: https://sandbox.stateset.app/  
Category: agent-sandbox · Isolation: gvisor

## Pricing regimes (raw)

- **On-demand** (resource): $0.128/vCPU-h, $0.0106/GiB-h
- **Team overage** (resource): $0.1/vCPU-h, $0.008/GiB-h

## Features

Yes: snapshots, fork/clone, persistent disk, custom image (Docker or snapshot), start from your own snapshot, preinstalled agents, HTTPS preview URLs, egress allowlist, private networking, self-hosting, SOC 2, SSO, Python, Node.js, credential injection, webhooks, audit logs, spend limits, automatic snapshots, snapshots on demand, agent harness API, gVisor or VM (no shared kernel), inbound access rules, secret proxy

No: memory snapshots, pause/resume, volumes, idle auto-stop, ≥24 h sessions, your own Docker/OCI image, full VM (own kernel), Docker inside, nested virtualization, root, browser, desktop GUI, computer-use API, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, SSH, public IPv4, static egress IP, BYOC, open source, HIPAA, GPU, arm64, Windows, macOS, HTTP method/path egress rules, wake on request, live resize, live fork (no pause), memory fork, EU data residency, MCP server, hosted agent API (their own agent), firewall inside (nftables), secret proxy for any API, extra volumes, shared volumes

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/stateset.json → feature_evidence

## Caveats

- Public homepage lists compute/network/create meters. gVisor Kubernetes pods, non-root. Free CPU-hours cannot be applied to RAM or network. Runtime price shown as CPU-hour; allocated versus observed CPU metering is not fully specified.
- ENGINE_CARD cannot apply all independent free CPU/RAM/storage/request allowances exactly. Raw regimes/worked example retain them; any engine subtotal may overstate charges or miss eligibility. Do not subtract them twice.
- Features set to null were not verified. No sandbox was purchased or tested; feature/performance statements are documentation claims. Null maximums are unknown unless a cited note explicitly states unlimited.
- A complete monthly bill requires known snapshot/egress/plan limits; do not silently treat missing ancillary charges as free.
- Plans re-checked 2026-09-28 on sandbox.stateset.app/pricing: Hobby $0 (10 concurrent, 45 min max), Pro $49 (2,000 concurrent, 5 h), Team $299 (10,000 concurrent, 5 h). Included vCPU-h / GB-h / network / creates are separate buckets kept as extra plan keys; the engine only understands a fungible included_usd, so they are not credited (small overstatement).
- Disk: no per-sandbox disk size or disk price on the pricing page; disk beyond 0 GiB stays unknown. Snapshots: checkpoints are filesystem + env only ("Process state is not captured"), max 10 GB, retained 90 days, stored in GCS/S3; no checkpoint storage price published (snapshot_gib_month null). The docs production guide carries an older conflicting table ($0.05/vCPU-h, $0.02/GB-h, $0.20/GB-month storage, different plan limits) that is not used.
- Access is API/SDK only (run commands, files, artifacts); docs mention no SSH, browser, desktop/VNC or computer-use offering, so those are set false. Pods run non-root with a read-only root filesystem (root false).
- Nested virtualization set to no: every sandbox runs under gVisor, which gives the workload its own user-space kernel and no /dev/kvm device (no KVM-based emulators or VMs inside). Docker inside is a separate question.

## How this provider charges


As of 2026-09-28. Prices are USD unless explicitly labelled otherwise. Source type: official published list/promo or unknown, never a fabricated quote.

Public homepage lists compute/network/create meters. gVisor Kubernetes pods, non-root. Free CPU-hours cannot be applied to RAM or network. Runtime price shown as CPU-hour; allocated versus observed CPU metering is not fully specified.

| Regime | When it applies | How billed / numbers | Source |
|---|---|---|---|
| Compute | Runtime | $0.128/CPU-hour + $0.011/GB-hour. | https://sandbox.stateset.app/ |
| Network | Metered transfer | $0.15/GB. | https://sandbox.stateset.app/ |
| Creates | Per create | $0.60/million = $0.0000006 each. | https://sandbox.stateset.app/ |

## Gotchas

- Public homepage lists compute/network/create meters. gVisor Kubernetes pods, non-root. Free CPU-hours cannot be applied to RAM or network. Runtime price shown as CPU-hour; allocated versus observed CPU metering is not fully specified.
- ENGINE_CARD cannot apply all independent free CPU/RAM/storage/request allowances exactly. Raw regimes/worked example retain them; any engine subtotal may overstate charges or miss eligibility. Do not subtract them twice.
- Features set to null were not verified. No sandbox was purchased or tested; feature/performance statements are documentation claims. Null maximums are unknown unless a cited note explicitly states unlimited.
- A complete monthly bill requires known snapshot/egress/plan limits; do not silently treat missing ancillary charges as free.

## Worked workload

4 vCPU / 8 GiB; 50 simultaneous instances × 8 h/day × 22 days = **8,800 instance-hours**, **35,200 vCPU-hours**, **70,400 GiB-hours**. At 30% observed CPU:10,560 active-vCPU-hours only where the meter actually uses utilization. Assume one start/instance/day:1,100starts. Retain50GiB snapshots for a month;100GiB outbound. Active disk capacity was not specified.

Under allocatedCPU assumption: compute$5,280 minus5×.128−420×.011=$5,274.74. Network(100−20)×.15=$12;1,100creates fit5,000free. Subtotal$5,286.74 before snapshots/storage and unknown plan/limit effects. Basis/concurrency/duration need confirmation.

## Sources

- https://sandbox.stateset.app/