Stekpad vs Firecrawl
Same five-verb shape. Different product. Here’s the honest comparison.
Pick the one that matches how you build.
- Stekpad is API-first scraping with built-in dataset storage, 19 native enrichers, and a cookie bridge that never stores your cookies. PAYG-first pricing, no subscription required.
- Firecrawl is API-first scraping that returns markdown by value. No native enrichment, no first-party cookie bridge, subscription pricing.
- Pick Stekpad if you want to keep, re-enrich and re-query your scraped data — or if your data must stay in one stack.
- Pick Firecrawl if you only need the markdown for a single agent call and you’re already integrated with their SDK.
24 rows of side-by-side.
Every meaningful difference, on the table. Not a marketing checklist.
| Feature | Stekpad | Firecrawl |
|---|---|---|
| Built-in dataset storage | Yes — every scrape lands in a persistent table or markdown bundle, re-queryable via API | No — returns by value, you store it yourself |
| Native enrichers | 19 enrichers in-house (find_emails, email_verify, company_enrich, geocode, ai_extract, ...) | 0 — bring your own |
| Cookie bridge (zero stored cookies) | Yes — Chrome extension fetches authenticated pages in your own browser | No — server-side cookies stored on their infra |
| First-class MCP server | Yes, 8 tools, shipped day one, reads free | Community MCP, no read tools, no per-call credit reporting |
| `scrape` verb | 1 credit, sync, formats: markdown / json / html / screenshot, 6 actions | 1 credit, sync, similar formats, similar actions |
| `crawl` verb | 1 credit / page, async, webhooks, dedupe by canonical URL, include/exclude paths | 1 credit / page, async, webhooks, similar |
| `map` verb | 1 credit / 1,000 URLs, sync, sitemap + robots + shallow link walk | Available, similar shape |
| `extract` verb (LLM) | 5 credits / URL, JSON Schema, LLM cascade Gemma → Haiku → Sonnet | Available, similar shape, single model |
| `search` verb | 5 credits + 1 / scraped result, optional in-call scrape | Available, similar shape |
| Webhook events | 11 events incl. row.added, row.changed, enrichment.completed, credits.low | ~5 events, fewer granular states |
| Batch extract | Async with run_id, polling + webhooks, 2 retries on schema fail | Sync only, smaller batches |
| Credit model | Single unified wallet, hard-stop at 0, no overage, failed runs auto-refund | Subscription credits, monthly reset |
| Team workspaces | Yes, from Cloud Growth (3 seats) and Cloud Scale (10 seats) | Available on enterprise tiers |
| SDK languages | TypeScript + Python in v1, Go + Rust community v1.5+ | TypeScript + Python + Go + Rust |
| MCP tool count | 8 (5 verbs + 3 read tools) | Community MCP, smaller surface |
| On-prem option | Cloud Scale + Enterprise | Available on enterprise |
| Self-host | Enterprise plan only, by contract | Open source self-host available |
| Pricing per 1,000 pages (`crawl`) | ~3 € on Pro pack (PAYG, no sub required) | ~16 $ on standard plan (subscription) |
| Data retention | 7 d (Free) / 30 d (PAYG) / 90 d (Cloud Starter) / 1 y (Growth) / unlimited (Scale) | Days to weeks, plan-dependent |
| Authenticated session support | Cookie bridge — cookies never leave your browser | Server-side cookie injection |
| schema.org / structured data extraction | Returned in `metadata.schema_org` automatically on every scrape | Manual extraction |
| Dedupe | By canonical URL (default), content hash, or custom primary_key array | By canonical URL |
| Rate-limit handling | Automatic backoff + structured `target_blocked` error with retry_after | Automatic backoff |
| Region coverage | EU + US datacenters in v1, residential proxies v1.5 | Global datacenter pool |
Four things that actually change how you build.
Storage is the product
Every Stekpad scrape lands in a re-queryable dataset with versioned rows and content-hash diffs. Firecrawl returns by value — you store it yourself.
19 native enrichers
Find emails, verify them, enrich companies, geocode, classify with an LLM — all in-house. Firecrawl ships zero enrichers; you stitch four vendors.
Cookie bridge, by design
Authenticated scraping where your cookies stay in your Chrome. There’s no place to store them on Stekpad servers. Firecrawl uses server-side cookies.
First-class MCP, day one
8 MCP tools, free dataset reads, typed errors, per-call credit reporting. Firecrawl ships a community MCP with a smaller surface.
The honest verdict.
We genuinely think there are situations where Firecrawl is the better choice. Be honest about them so we don’t waste anyone’s time.
- You only need a single markdown blob and you’re already integrated with their SDK. Switching costs are real.
- You want a self-hosted open-source crawler today. Firecrawl ships an OSS version. Stekpad’s self-host path is enterprise-only.
- You need Go or Rust SDKs in production right now. Firecrawl ships them officially in v1; Stekpad’s are community projects until v1.5.
- You’re scraping 100% public, anonymous content with no need to keep it. Stekpad’s dataset model is overhead in that case — stick with what’s working.
Where we win on substance.
- You want every scrape to live in a re-queryable dataset — for change monitoring, re-enrichment, or just remembering what you scraped last week.
- You need authenticated scraping without storing cookies on a third-party server. The cookie bridge is the only honest path there.
- You want enrichment in the same product — emails, phones, companies, socials, geocodes — without stitching three vendors.
- You’re building an AI agent on Claude/Cursor/Claude Code. The first-class MCP server and free dataset reads are designed for this.
- You prefer pay-as-you-go credits that last 12 months over a subscription that resets every month.
- You care about data residency — your scraped data never leaves Stekpad’s stack.
Common questions.
Can I migrate from Firecrawl to Stekpad?
Yes — the verb shapes are similar enough that most migrations are a one-day SDK swap. Your stored datasets in Firecrawl don’t transfer (they don’t have any), so there’s no migration of historical data.
Does Stekpad support every Firecrawl format?
Yes — markdown, JSON (via schema), HTML, screenshot. Same four output shapes.
Why doesn't Stekpad have a self-hosted open-source version?
Self-hosting a scraper at scale is an infrastructure problem, not a software problem. We’d rather offer a real cloud product with honest pricing and a path to dedicated infra on Enterprise than ship a hard-to-operate OSS edition.
Is the comparison up to date?
We update this page when either side ships a material change. Last updated alongside the v3 launch. Spot something stale? Email josua@stekpad.com.
Can I see both products side-by-side on a real workload?
Yes — pick a docs site, run the same `crawl` on both, compare the markdown quality and the time to "where do I store this". That’s the honest test.
Try the Stekpad API.
300 free credits a month, no subscription, MCP server included.