# Nodus Compute

Official pricing: https://www.nodus-compute.ai/pricing/  
Category: agent-sandbox · Isolation: not published

## Pricing regimes (raw)

- **Admitted managed sandbox** (resource) [beta]

## Features

Yes: snapshots, pause/resume, persistent disk, volumes, idle auto-stop, ≥24 h sessions, custom image (Docker or snapshot), start from your own snapshot, your own Docker/OCI image, SSH, HTTPS preview URLs, egress allowlist, BYOC, GPU, Python, Node.js, wake on request, MCP server, automatic snapshots, hosted agent API (their own agent), inbound access rules, secret proxy, extra volumes

No: memory snapshots, fork/clone, Docker inside, nested virtualization, root, browser, desktop GUI, computer-use API, browser + desktop control, anti-bot stealth, CAPTCHA solving, residential IPs, preinstalled agents, public IPv4, raw TCP inbound, static egress IP, self-hosting, open source, SOC 2, HIPAA, SSO, arm64, Windows, macOS, HTTP method/path egress rules, live resize, live fork (no pause), memory fork, EU data residency, snapshots on demand, agent harness API, firewall inside (nftables), secret proxy for any API, shared volumes

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

## Caveats

- Published table says supply costs, NOT final all-in customer tariffs. Live meter separates compute/platform/storage/model/subscription charges. Do not rank the $0.0714 4/8 supply figure as a customer price.
- CPU managed agent tools sandbox is gated by account admission and deployment qualification. Final billing can settle after completion; retained files keep billing. Legacy per-run budget fields do not enforce spend limits.
- Sleep saves a verified project and releases compute; wake rebuilds tools/processes, not proof of memory snapshot.
- Null values mean unknown, not free/unlimited. No model-token cost, tax or unverified ancillary charge included.
- Native browser/device/task/credit prices with unknown CPU/RAM are deliberately non-priceable. Check regime before interpreting any partial resource estimate.
- Re-checked 2026-09-28: nodus-compute.ai/pricing still labels its CPU/GPU table "supply costs", not end-customer prices (e.g. 4 vCPU/16 GiB/64 GiB disk $0.1054/h supply); docs/concepts/costs gives no customer rate card. Kept unpriced.
- Pricing page unchanged since 2026-09-28: still titled 'GPU and CPU supply costs.' with a 'Nodus supply' column next to Modal list prices; no customer rate card. The meter adds separate platform fees ('The meter separates compute, platform fees, storage, model and subscription charges'). Keep unpriced/beta.
- Admission-gated preview: CPU agent sandboxes, managed agents and Workspaces all require 'account admission and deployment qualification'. Terms: 'Pilot / invite access may be limited or revoked' and 'During launch and pilot periods, no SLA applies'.
- No per-run spend caps: legacy budget fields are ignored; work stops only when credits run out. Managed agents default to 0-4 workers; 'Requested warm workers can incur idle charges.'
- GPU Workspaces (preview, 2026-09-24): saved project + GPU machine with browser editor/VS Code over SSH/JupyterLab; compute billed while running, saved files kept between sessions; no published customer rate.

## How this provider charges


YC: Fall 2026; directory status **Active**. Research scope: public-product-researched. Source: https://www.ycombinator.com/companies/nodus-compute

Intelligent Cloud for AI workloads

## Regime table

| Regime | When it applies | How billed | Numbers | Source |
|---|---|---|---|---|
| Admitted managed sandbox | Final customer rates unknown; supply quotes are not all-in bill. | resource; CPU alloc, memory alloc | Unknown or native-unit billing; see facts below | https://www.nodus-compute.ai/pricing/ |
| Unverified | Plan limits apply | Monthly fee; credits only as stated | fee=None; included usage=$None; concurrent=None; max session h=None | https://www.nodus-compute.ai/pricing/ |
| Retained disk / GiB-month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://www.nodus-compute.ai/pricing/ |
| Snapshots / GiB-month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://www.nodus-compute.ai/pricing/ |
| Egress / GiB | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://www.nodus-compute.ai/pricing/ |
| IPv4 / month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://www.nodus-compute.ai/pricing/ |

## Verified billing facts and gotchas

1. Published table says supply costs, NOT final all-in customer tariffs. Live meter separates compute/platform/storage/model/subscription charges. Do not rank the $0.0714 4/8 supply figure as a customer price.
2. CPU managed agent tools sandbox is gated by account admission and deployment qualification. Final billing can settle after completion; retained files keep billing. Legacy per-run budget fields do not enforce spend limits.
3. Sleep saves a verified project and releases compute; wake rebuilds tools/processes, not proof of memory snapshot.

Native pricing (not coerced into incompatible engine units):
```json
{
  "cpu_supply_sizes": [
    {
      "name": "1/1",
      "vcpu": 1,
      "ram_gib": 1,
      "disk_gib": 25,
      "hour": 0.0075,
      "month_cap": null
    },
    {
      "name": "1/2",
      "vcpu": 1,
      "ram_gib": 2,
      "disk_gib": 50,
      "hour": 0.0179,
      "month_cap": null
    },
    {
      "name": "2/4",
      "vcpu": 2,
      "ram_gib": 4,
      "disk_gib": 80,
      "hour": 0.0357,
      "month_cap": null
    },
    {
      "name": "4/8",
      "vcpu": 4,
      "ram_gib": 8,
      "disk_gib": 160,
      "hour": 0.0714,
      "month_cap": null
    },
    {
      "name": "2/8",
      "vcpu": 2,
      "ram_gib": 8,
      "disk_gib": 64,
      "hour": 0.0558,
      "month_cap": null
    },
    {
      "name": "4/16",
      "vcpu": 4,
      "ram_gib": 16,
      "disk_gib": 64,
      "hour": 0.1054,
      "month_cap": null
    }
  ],
  "gpu_supply_hour_usd": {
    "H100 SXM": 2.6,
    "A100-80GB": 1.09,
    "L40S": 0.793,
    "H200 SXM": 3.593,
    "RTX PRO 6000": 1.35,
    "RTX4090": 0.29,
    "RTX3090": 0.18,
    "RTX5090": 0.48,
    "A100-40GB": 0.904,
    "L4": 0.43,
    "RTX A6000": 0.333,
    "RTX6000 Ada": 0.743,
    "RTX A4000": 0.173,
    "RTX3070": 0.133,
    "B200": 5.5,
    "B300": 6.943
  }
}
```

## Worked example

Requested: 4 vCPU/8 GiB, 50 simultaneous × 8 h/day × 22 days = **8,800 instance-hours**, 30% CPU; 50 GiB snapshots and 100 GiB egress. A month is 730 hours only when explicitly used in rate conversion.

4/8 supply-only arithmetic: 8,800×$0.0714=$628.32. This is NOT a bill estimate: platform fees, retained storage, egress and admission terms are unverified. 30% CPU is not shown to reduce supply charges.

## Sources and coverage

- https://www.ycombinator.com/companies/nodus-compute
- https://www.nodus-compute.ai/pricing/
- https://www.nodus-compute.ai/docs/concepts/costs/
- https://www.nodus-compute.ai/docs/guides/agent-sandboxes/
- https://www.nodus-compute.ai/
- https://www.nodus-compute.ai/docs/
- https://www.nodus-compute.ai/llms.txt

Page captures and documentation indexes/bundles are in `raw/`; provider JSON lists the evidence paths. HN query results were captured separately as `<id>--hn.txt`; they are discovery leads, not current tariff authority. Downloaded docs do not imply every feature was verified; unsupported assertions remain null.