# TinyFish

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

## Pricing regimes (raw)

- **Native product meter; not a general-purpose 4-vCPU/8-GiB VM** (sizes):  [alt]
- **Monitor (Page/Topic change tracking)** (sizes):  [alt]

## Features

Yes: idle auto-stop, ≥24 h sessions, browser, browser automation API, anti-bot stealth, residential IPs, open internet, BYOC, SSO, Python, Node.js, webhooks, scheduled runs (cron), audit logs, MCP server, hosted agent API (their own agent)

No: snapshots, memory snapshots, fork/clone, pause/resume, persistent disk, volumes, start from your own snapshot, your own Docker/OCI image, Docker inside, nested virtualization, desktop GUI, computer-use API, desktop control API, browser + desktop control, CAPTCHA solving, preinstalled agents, SSH, public IPv4, HTTPS preview URLs, egress allowlist, static egress IP, self-hosting, open source, SOC 2, HIPAA, GPU, arm64, Windows, macOS, 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/tinyfish.json → feature_evidence

## Caveats

- A browser has no published vCPU/RAM allocation; $0.12/browser-hour is not a 4/8 VM.
- Default browser concurrency 5 cannot meet 50; custom sales limits required.
- Model inference is included with Agent steps; do not add it again.
- Per-minute price does not establish rounding increment or minimum session bill.
- Existing Agent run finishes when Wallet reaches zero, so zero balance is not an immediate kill guarantee.
- Native billing cannot be reduced to a 4-vCPU/8-GiB price. No matching engine sizes, by design.
- Agent runs default to max 150 steps (1-500 via beta-gated agent_config.max_steps) and a 10-minute run timeout; worst-case default Agent run = 150 x $0.016 = $2.40.
- Wallet can go negative: an in-flight Agent run finishes at $0 and new starts return 402 until money is added (example balance $-2.50).

## How this provider charges


As of 2026-09-28. Public USD list rates; prepaid Wallet. Browser capacity and Agent actions are distinct meters. Show HN 2026-02-12 item 46991520.

## Regimes

| Regime | When it applies | How billed / numbers | Source |
|---|---|---|---|
| Browser | Native product usage | $0.002/browser-minute = $0.12/browser-hour; 5 default concurrent. | [Primary pricing](https://www.tinyfish.ai/pricing) |
| Agent | Native product usage | $0.016/step; 2 default concurrent; LLM included. | [Primary pricing](https://www.tinyfish.ai/pricing) |
| Search / Fetch | Native product usage | Free within Search 30/min and 500/hour; Fetch 150 URLs/min and 1,000/day. Different products, not free browsers. | [Primary pricing](https://www.tinyfish.ai/pricing) |
| Included infrastructure | Native product usage | Residential proxies, anti-bot handling, screenshot/log/source storage included; duration/size terms not established. | [Primary pricing](https://www.tinyfish.ai/pricing) |
| Pay as you go | Account plan | $0/month; Browser 5 concurrent; Agent 2 concurrent; no subscription/minimum monthly spend. | [Primary pricing](https://www.tinyfish.ai/pricing) |
| Enterprise | Account plan | Unpublished / sales; Custom concurrency and rate limits; VPC, SSO, audit logs and contractual SLA. | [Primary pricing](https://www.tinyfish.ai/pricing) |

## Gotchas

- A browser has no published vCPU/RAM allocation; $0.12/browser-hour is not a 4/8 VM.
- Default browser concurrency 5 cannot meet 50; custom sales limits required.
- Model inference is included with Agent steps; do not add it again.
- Per-minute price does not establish rounding increment or minimum session bill.
- Existing Agent run finishes when Wallet reaches zero, so zero balance is not an immediate kill guarantee.

## Required worked example

4 vCPU / 8 GiB, 50 concurrent × 8 h/day × 22 days = **8,800 instance-hours**, 30% CPU, 50 GiB retained snapshots and 100 GiB egress.

**No valid VM total.** A separate browser-only arithmetic scenario is `8,800 × $0.12 = $1,056` before any contract changes, but 50 concurrency exceeds the public limit of 5. $1,056 is neither a sales quote nor a 4/8-VM estimate. Agent steps, snapshot retention and general egress cannot be inferred from these hours.

## Feature evidence

- https://docs.tinyfish.ai/llms-full.txt
- https://docs.tinyfish.ai/browser-api
- https://docs.tinyfish.ai/agent-api