How is TutuSoo different from Firecrawl?

Firecrawl leans toward "scrape a given page to Markdown"; TutuSoo leans toward "retrieve a set of relevant results from the whole web". For RAG / grounded Q&A the latter is less work.

Does it handle Chinese well?

Yes, and it is tuned for it: the index uses jieba tokenization, so Chinese recall is clearly better than treating a phrase as one token.

Can I see where results come from?

Yes. Every result carries a source tag (local index or brave fallback) so you can judge trust and cite.

Is there rate limiting?

Yes. With a key it counts per-key rate_limit with X-RateLimit-* headers; without a key it uses the public per-IP tier.

Firecrawl alternative — the TutuSoo API

Own BM25 full-text index + web fallback, structured results from one request.

Read the search API docs

Firecrawl is great at scraping a single page into LLM-friendly Markdown. If what you need is not "scrape this page" but "ask the whole web and get back a set of relevant results", TutuSoo fits better.

TutuSoo runs its own crawler + Tantivy BM25 full-text index, falls back to external search when local results are thin, and returns structured JSON with title / snippet / source (local/brave) — real-time retrieval for RAG, agents and AI apps, with llms.txt support.

TutuSoo vs Firecrawl

對比項	TutuSoo	Firecrawl
Core capability	Web-wide retrieval, a set of results	Scrape a given page to Markdown
Own index	✅ Tantivy BM25 full-text	Mostly live scraping
CJK tokenizer	✅ jieba	Generic
Source attribution	✅ local / brave	—
llms.txt	✅ Supported	Partial

主要功能

Own web index

spider-rs crawler + Tantivy BM25 index focused on high-authority sites — not just single-page scraping.

Structured JSON

GET /api/v1/search returns title / url / snippet / source in one request, ready for your model.

Strong CJK search

title/body tokenized with jieba, so Chinese queries recall properly instead of one giant token.

Safe & compliant

Respects robots.txt + Crawl-delay; outbound SSRF guard rejects private / loopback / cloud-metadata.

使用步驟

1
Get an API key

Request a key on the developer page, or try the public IP-based tier for free first.
2
Make one request

GET /api/v1/search?q=your+question with your Authorization header.
3
Use the results

Parse the results array (title/url/snippet/source) and feed it to your LLM or render directly.

常見問題

How is TutuSoo different from Firecrawl?: Firecrawl leans toward "scrape a given page to Markdown"; TutuSoo leans toward "retrieve a set of relevant results from the whole web". For RAG / grounded Q&A the latter is less work.
Does it handle Chinese well?: Yes, and it is tuned for it: the index uses jieba tokenization, so Chinese recall is clearly better than treating a phrase as one token.
Can I see where results come from?: Yes. Every result carries a source tag (local index or brave fallback) so you can judge trust and cite.
Is there rate limiting?: Yes. With a key it counts per-key rate_limit with X-RateLimit-* headers; without a key it uses the public per-IP tier.

現在就接入圖圖搜 API

免費額度 · 一行請求即得結構化結果 · 支援 llms.txt

Read the search API docs