# H Company / Surfer / H Agents API

Official pricing: https://www.hcompany.ai/pricing  
Category: browser · Isolation: not published

## Pricing regimes (raw)

- **Managed H Agents browser** (sizes): Managed H Agents browser ? vCPU/? GiB $null/h

## Features

Yes: pause/resume, idle auto-stop, ≥24 h sessions, browser, computer-use API, browser automation API, residential IPs, open internet, static egress IP, self-hosting, open source, SOC 2, webhooks, scheduled runs (cron), EU data residency, MCP server, hosted agent API (their own agent)

No: snapshots, memory snapshots, fork/clone, persistent disk, volumes, start from your own snapshot, your own Docker/OCI image, Docker inside, nested virtualization, desktop GUI, desktop control API, browser + desktop control, anti-bot stealth, CAPTCHA solving, preinstalled agents, SSH, public IPv4, HTTPS preview URLs, egress allowlist, BYOC, GPU, arm64, Windows, macOS, Python, Node.js, HTTP method/path egress rules, wake on request, live resize, live fork (no pause), memory fork, automatic snapshots, snapshots on demand, agent harness API, inbound access rules, firewall inside (nftables), secret proxy, secret proxy for any API, extra volumes, shared volumes

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/h-company.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.
- Quoted session counts are estimates dependent on model and task complexity, not fixed session-hour entitlements. Credits cover model work, not a normalized resource-hour rate.
- Current hosted product is H Agents API. Local desktop support must not be represented as a hosted Windows/macOS VM SKU; docs distinguishes cloud browser from own desktop.
- Holo 3.1 free model tier 10 RPM appears on Models tab; not the same as free hosted agent concurrency. Model rates captured with actual tab click.
- CONFLICT: the developer docs' plan table does not match the marketing pricing page. Docs list Free $0 (15,000,000 tokens/period, 3 concurrent) and Developer $29/month (65,000,000 tokens/period, 10 concurrent). The pricing page lists Free / Builder $20 / Startup $200 / Enterprise with model-cost credit and +20% top-ups. The 2026-09-22 changelog ('included agent usage', 'Top up credit') matches the pricing page, so the docs table is probably stale. The card follows the pricing page.
- Above the concurrency cap, sessions are queued by default (they start oldest-first as slots free) instead of being rejected; queue:false restores 429. Concurrency caps delay bursts rather than failing them.
- Out of credit: sessions end failed with error_code insufficient_credit (changelog 1.0.11, 2026-09-22). No automatic overage is documented; you must top up.
- Session data (events, screenshots) is kept 30 days by default, configurable (delete_after_min, null = indefinitely); no storage price published.
- Only the browser is hosted by H; desktop control is on the user's own machine. Browser options: lazy start (a browser the agent never uses is never started), headless, visual viewport default 1200x1200, persistent/default browser profiles, managed residential/datacenter proxy (no per-GB price), cron schedules.
- hcompany.ai/surfer-h now serves the 'Surfer 2' article; the old Surfer-H $0.13/task benchmark figure is no longer published (historical only, never used as a price).

## 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 |
|---|---|---|---|---|
| Hosted Agents API | Managed cloud browser | Included model dollars then 20% markup | Builder $20 /10 concurrency, Startup $200 /100 concurrency | https://www.hcompany.ai/pricing |
| Models API | Bring environment | Tokens | Input/output per million: .40/3; .30/2; .25/1.80; .40/3 across displayed models | https://www.hcompany.ai/pricing |
| Local desktop/browser | User machine | No cloud VM included | Holo Desktop CLI, HoloTab, open weights; inference separate | https://www.hcompany.ai/pricing |
| Surfer-H benchmark | Research evaluation | Not a purchasable tariff | $0.13/task was benchmark cost; do not use as cloud list price | https://www.hcompany.ai/surfer-h |
| Free | Account plan | Monthly fee / commitment | $0; concurrency 3; ~60 sessions/month, token fair-use cap, datacenter proxy. | https://www.hcompany.ai/pricing |
| Builder | Account plan | Monthly fee / commitment | $20; concurrency 10; First $20 at model cost (~1,000 illustrative sessions); top-ups model cost +20%; residential proxy. | https://www.hcompany.ai/pricing |
| Startup | Account plan | Monthly fee / commitment | $200; concurrency 100; First $200 at model cost (~10,000 illustrative sessions); top-ups +20%; residential proxy. | https://www.hcompany.ai/pricing |
| Enterprise | Account plan | Monthly fee / commitment | Quote / unknown; concurrency unpublished/custom; Negotiated markup/token cost, uncapped/custom concurrency, workflow orchestration. | https://www.hcompany.ai/pricing |

## Gotchas

- Quoted session counts are estimates dependent on model and task complexity, not fixed session-hour entitlements. Credits cover model work, not a normalized resource-hour rate.

- Current hosted product is H Agents API. Local desktop support must not be represented as a hosted Windows/macOS VM SKU; docs distinguishes cloud browser from own desktop.

- Holo 3.1 free model tier 10 RPM appears on Models tab; not the same as free hosted agent concurrency. Model rates captured with actual tab click.

- 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": null,
  "included_hours": null,
  "proxy_residential_gb_usd": null,
  "proxy_datacenter_gb_usd": null,
  "captcha_solve_usd": null,
  "stealth_surcharge_usd": null,
  "recording_storage_gib_month_usd": null,
  "recording_retention": null,
  "persistent_profiles": true,
  "resolution": null,
  "gpu_desktops": null,
  "os_options": [],
  "browsers": [
    "Cloud browser; local Chrome alternative"
  ]
}
```

## 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 simultaneous managed agents fits Startup ($200 monthly entry). Total depends on model token usage T: first $200 at model cost, then approximately $200+1.2×max(0, T−200), subject to plan terms. 8,800 browser-hours provides no T. Do not multiply hours by historical $0.13 benchmark task cost; machine sizing/storage/network unknown.

## Sources

- https://www.hcompany.ai/pricing

- https://hub.hcompany.ai/llms-full.txt

- https://www.hcompany.ai/surfer-h