# Kernel

Official pricing: https://www.onkernel.com/pricing  
Category: browser · Isolation: firecracker

## Pricing regimes (raw)

- **Headless** (sizes): Headless ? vCPU/1 GiB $0.06000012/h
- **Headful (8 or 16 GiB)** (sizes): Headful 8 GiB ? vCPU/8 GiB $0.48000096/h; Headful 16 GiB ? vCPU/16 GiB $0.96000192/h
- **Headful GPU** (sizes): GPU ? vCPU/? GiB $2.88000576/h
- **App invocation (serverless agent code, 4 GB)** (resource): $0/vCPU-h, $0.06000012/GiB-h [alt]

## Features

Yes: snapshots, memory snapshots, pause/resume, idle auto-stop, ≥24 h sessions, custom image (Docker or snapshot), start from your own snapshot, full VM (own kernel), root, browser, desktop GUI, computer-use API, browser automation API, desktop control API, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, SSH, egress allowlist, open internet, static egress IP, BYOC, open source, SOC 2, HIPAA, SSO, GPU, Python, Node.js, wake on request, audit logs, EU data residency, zero data retention, spend limits, MCP server, automatic snapshots, gVisor or VM (no shared kernel), inbound access rules, firewall inside (nftables)

No: fork/clone, persistent disk, volumes, your own Docker/OCI image, Docker inside, nested virtualization, public IPv4, HTTPS preview URLs, self-hosting, Windows, macOS, HTTP method/path egress rules, live resize, live fork (no pause), memory fork, snapshots on demand, agent harness API, hosted agent API (their own agent), secret proxy, secret proxy for any API, extra volumes, shared volumes

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

## Caveats

- Storage, snapshot GiB-month and public IPv4 prices are not published in the reviewed sources; null is not free.
- No guaranteed vCPU allocation published; browser-hour is not a verified 4-vCPU/8-GiB VM equivalent. The sizes card keeps unknown hardware null; do not coerce it to zero or auto-rank against resource-guaranteed VMs.
- Browser RAM is not published unless a specific mode overrides it.
- The homepage GB-second rate and docs preset rates agree: 1 GB headless, 8 GB headful. 16 GB is a documented headful option. Do not charge CPU+RAM again on top.
- Standby billing is based on browser activity, not CPU utilization. An open active live view/CDP workload may prevent savings. Pool and standby browsers continue consuming concurrency.
- Pool availability conflicts internally: feature table says all tiers; FAQ says Start-Up and Enterprise. Fifty-browser example already needs Start-Up.
- Idle pool disks are free, but snapshot_gib_month stays null because general snapshot storage is not priced. Browser profiles are not general disks.
- Headful = Chromium on a virtual display with Live View and a computer-controls API, not a full desktop environment; SSH into the browser VM is available (kernel browsers ssh). Headless has no Live View.
- eu-west / ap-southeast browsers need the Start-Up or Enterprise plan. On Free and Hobbyist Kernel is US-only (us-east), so an EU/Asia workload must pay the USD 200 Start-Up plan. Model this as region-specific mode copies with plans:["Start-Up","Enterprise"], or as a plan gate.
- 16 GiB headful price (USD 0.96/h) is derived from the GB-second definition; the docs pricing table lists only one headful rate (USD 0.0001333336/s = 8 GB).
- YC startups get a discount ('here's our yc deal'). Terms are behind a link and not published on the pricing page.

## How this provider charges


As of 2026-09-28. USD unless explicitly stated; public list prices unless a row is marked promotion, historical, or quote-only.

## Regime table

| Regime | When it applies | How billed | Numbers / terms | Source |
|---|---|---|---|---|
| Browser runtime | Active headless/headful/GPU | Per second | $0.0000166667 / $0.0001333336 / $0.0008000016 per second | https://kernel.sh/docs/info/pricing |
| Standby | 5 seconds without CDP or live-view activity | No runtime charge | $0 but concurrency slot held | https://kernel.sh/docs/info/pricing |
| Pools | Idle prewarmed capacity | Free idle runtime/disk | Reserved pool size counts toward concurrency even unacquired | https://kernel.sh/docs/info/pricing |
| Proxy / stealth / CAPTCHA | Managed service | Included | $0 extra; proxy configuration plan-gated | https://kernel.sh/docs/info/pricing |
| App runtime | Serverless agent execution | Active code runtime | $0.24000048/hour plus browsers and LLMs | https://kernel.sh/docs/info/pricing |
| Free / Developer | Account plan | Monthly fee / commitment | $0; concurrency 5; 1-day replays, 3 auth connections and vaults. | https://www.onkernel.com/pricing |
| Hobbyist | Account plan | Monthly fee / commitment | $30; concurrency 10; 7-day replays; configured proxies, unlimited auth/vaults. | https://www.onkernel.com/pricing |
| Start-Up | Account plan | Monthly fee / commitment | $200; concurrency 150; GPU, regional browsers and BYO proxy; 30-day replays. App concurrency 50 org /20 per app. | https://www.onkernel.com/pricing |
| Enterprise | Account plan | Monthly fee / commitment | Quote / unknown; concurrency unpublished/custom; SSO, HIPAA BAA, dedicated support, ZDR, custom limits. | https://www.onkernel.com/pricing |

## Gotchas

- The homepage GB-second rate and docs preset rates agree: 1 GB headless, 8 GB headful. 16 GB is a documented headful option. Do not charge CPU+RAM again on top.

- Standby billing is based on browser activity, not CPU utilization. An open active live view/CDP workload may prevent savings. Pool and standby browsers continue consuming concurrency.

- Pool availability conflicts internally: feature table says all tiers; FAQ says Start-Up and Enterprise. Fifty-browser example already needs Start-Up.

- Idle pool disks are free, but snapshot_gib_month stays null because general snapshot storage is not priced. Browser profiles are not general disks.

- Storage, snapshot GiB-month and public IPv4 prices are not published in the reviewed sources; null is not free.

- No guaranteed vCPU allocation published; browser-hour is not a verified 4-vCPU/8-GiB VM equivalent. The sizes card keeps unknown hardware null; do not coerce it to zero or auto-rank against resource-guaranteed VMs.

- Browser RAM is not published unless a specific mode overrides it.

## Browser and desktop dimensions

```json
{
  "hour_usd": 0.06,
  "included_hours": null,
  "proxy_residential_gb_usd": 0,
  "proxy_datacenter_gb_usd": 0,
  "captcha_solve_usd": 0,
  "stealth_surcharge_usd": 0,
  "recording_storage_gib_month_usd": null,
  "recording_retention": "1/7/30 days Free/Hobbyist/Start-Up",
  "persistent_profiles": true,
  "resolution": "1920×1080 at 25 Hz default; configurable; refresh rate constrained by pixels",
  "gpu_desktops": "Headful GPU $2.88000576/hour, Start-Up+, GPU model unknown",
  "os_options": [
    "linux"
  ],
  "browsers": [
    "Chromium"
  ]
}
```

## Worked example

Requested: 4 vCPU / 8 GiB; 50 concurrent × 8 hours/day × 22 days = **8,800 runtime hours**, 1,100 eight-hour starts if recreated daily; 30% CPU; 50 GiB-month snapshots; 100 GiB egress.

50 concurrency requires Start-Up. At 8,800 fully active hours: headless $200+max(0,528.001056−50)=**$678.001056**; 8 GiB headful **$4,374.008448**; GPU **$25,494.050688**. Managed proxies have no per-GB charge. These are native browser subtotals, not guaranteed 4-vCPU machines. If a measured fraction a enters paid active runtime, replace hours with 8,800 a; 30% CPU alone cannot set a. General 50 GiB snapshots are not priced.

## Sources

- https://www.onkernel.com/pricing

- https://docs.kernel.sh/llms-full.txt

- https://kernel.sh/docs/info/pricing

- https://kernel.sh/docs/browsers/standby

- https://kernel.sh/docs/browsers/headless

- https://kernel.sh/docs/browsers/viewport

- https://kernel.sh/docs/browsers/processes