Skip to main content New gottem — one API for every web scraping vendor.
Spider Browser

The browser your agents won't get blocked on.

Pre-warmed sessions, stealth on by default, CAPTCHA handling, and proxy rotation behind a single connection. Drop the SDK into your agent, or run the desktop app — zero infrastructure, ~4ms latency.

Launch browser Read the docs
agent.ts Live
import { SpiderBrowser } from "spider-browser"

const browser = new SpiderBrowser({
  apiKey: process.env.SPIDER_API_KEY!,
  mode: "cua",
})
await browser.init()
await browser.page.goto("https://example.com")

// AI agent completes tasks autonomously
const result = await browser
  .agent({ maxRounds: 20 })
  .execute("Find the cheapest flight to Tokyo")

await browser.close()
agent complete · 8 steps · $0.014
stealth bench
#1
command latency
~4ms
cold start
0ms
01 · Why this one

Not a Chrome wrapper.

A purpose-built Rust browser engine, sized for agents and scrapers — not bolted on top of someone else's browser stack.

01

Sub-millisecond commands

Custom Rust engine, no Chrome overhead, no Node.js bottleneck. The first command after `init()` runs in microseconds — not seconds.

02

Stealth on by default

Fingerprint randomization, proxy rotation, and CAPTCHA solving are wired in. Stealth escalates automatically if the target pushes back.

03

Pay for what runs

$1 per GB of bandwidth and $0.001 per minute of compute. No monthly minimum, no seat fees. Stealth and proxies are included in the price.

Rust engine Zero cold start Anti-detect Stealth Proxy rotation CAPTCHA solving AI extraction Screenshots Automation Session replay Structured data
02 · Stealth Bench V1

Your agents don't get blocked.

Spider scored 84.5% across 71 anti-bot tasks — the highest of any cloud browser tested. Browserbase scored 41.9%. Local headless Chrome scored 2.2%.

Stealth Bench V1: Spider leads with 84.5% success rate across 71 anti-bot tasks.Spider84.5%Browser Use81.4%Anchor76.8%Onkernel67.1%Browserless54%Steel47.2%Browserbase41.9%Hyperbrowser39.9%
Spider · 84.5%
Other cloud browsers
Methodology on GitHub ↗
03 · How it works

Install, drive, ship.

01

Connect

Install the SDK and call init(). Spider provisions a pre-warmed browser and hands you a live session in milliseconds.

02

Interact

Drive the browser with familiar APIs — goto(), act(), observe(), extract(). Stealth, proxies, and retries are handled.

03

Ship

Receive structured JSON, screenshots, or session recordings. Close the session and pay only for what you used.

04 · What you build

One API for the whole workflow.

Scraping

Web scraping at scale

Scrape any site, even behind anti-bot protection. Stealth and proxies are automatic — point it at a URL and get clean data.

Extraction

Structured extraction

Pull product listings, pricing, contacts, or any structured field with natural-language prompts or typed schemas.

Automation

Browser automation

Automate logins, form submissions, and multi-step flows. Plain-English commands or standard browser APIs — no fragile selectors.

Monitoring

Price and change monitoring

Track prices, inventory, or content changes across many sites. Smart retry means you get the data even from unreliable pages.

AI agents

Real browser for agents

Give an LLM a real browser via the agent() API. It can browse, click, fill forms, and complete multi-step tasks on its own.

Pipelines

Feed your data pipeline

Ingest JS-heavy pages into RAG, data warehouses, or search indexes. Spider renders SPAs and returns clean, structured content.

05 · SDKs

Your language, same API.

Official SDKs for TypeScript, Python, and Rust. Identical surface across all three.

TypeScript

npm ↗
$npm install spider-browser
import { SpiderBrowser } from "spider-browser"

const b = new SpiderBrowser({ apiKey })
await b.init()
await b.page.goto("https://example.com")
const data = await b.extract("Get prices")

Python

PyPI ↗
$pip install spider-browser
from spider_browser import SpiderBrowser

async with SpiderBrowser(api_key=key) as b:
    await b.page.goto("https://example.com")
    data = await b.extract("Get prices")
$cargo add spider-browser
use spider_browser::SpiderBrowser;

let b = SpiderBrowser::new(api_key)
    .init().await?;
b.goto("https://example.com").await?;
let data = b.extract("Get prices").await?;
06 · Spider vs the rest

Skip the infrastructure.

FeatureSpiderSelf-hostedOther clouds
Stealth Bench V1#1 (84.5%)N/A≤81%
Stealth & anti-botBuilt-inDIYPartial
CAPTCHA solvingAutomaticManualAdd-on
Smart retryAutoCustomLimited
Custom Rust engineYesN/ANo
AI extractionNativeNoneNone
Command latency~4ms~50ms~20ms
Cold start0ms3–10s1–5s
Session recordingBuilt-inCustomExtra cost
07 · FAQ

Common questions.

What is Spider Browser?

A cloud browser built from the ground up in Rust — not a Chrome wrapper. Sessions are pre-warmed, so they start in milliseconds, and stealth, proxy rotation, CAPTCHA solving, and smart retry are on by default. Use it for scraping, automation, extraction, or AI-agent workflows.

How is it different from Puppeteer or Playwright?

Puppeteer and Playwright drive Chrome — never designed for scraping or agents. Spider Browser is a purpose-built Rust engine: no Chrome overhead, no Node.js bottleneck, no DIY stealth. You write the logic, Spider handles the rest.

What can I use it for?

Anything that needs a real browser: scraping at scale, price monitoring, lead generation, competitive intelligence, automated testing, AI-agent browsing, RAG ingestion. JavaScript and login flows are first-class.

Which languages are supported?

Official SDKs for TypeScript (npm install spider-browser), Python (pip install spider-browser), and Rust (cargo add spider-browser). All three share the same API surface — init, goto, act, observe, extract, agent.

How does anti-bot bypass work?

Spider scored 84.5% on Stealth Bench V1, #1 across all cloud browsers tested. Stealth escalates automatically on failure; CAPTCHA solving and proxy rotation are built in. No configuration required.

What does it cost?

$1 per GB of bandwidth and $0.001 per minute of compute. No monthly minimum, no contracts. 500 free credits on sign-up. Stealth, proxies, and CAPTCHA solving are included in the price.

How many concurrent sessions can I run?

Up to 100 concurrent sessions with zero cold start. Each runs on a pre-warmed instance, so combined with smart retry and automatic recovery it scales to high-volume workloads on heavily protected sites.

Can AI agents use it?

Yes — the agent() API lets any LLM browse, click, fill forms, and complete multi-step tasks autonomously. act(), observe(), and extract() give fine-grained control with plain-English commands.

Start automating

Ship the agent in minutes.

2,500 free credits on sign-up. No credit card, no infrastructure, no cold starts.