Compared with Tavily (tavily.com)
The Tavily alternative, 5.7× cheaper
Search built for agents, like Tavily — 5.7× cheaper per search, with a calibrated relevance probability on every result and the top ones read and verified before they’re ranked. One request to rewrite; your agent stays the same.
Same job, 5.7× cheaper
Search
$1.40
Tavily: $8−83%
Deep search
$2.40
Tavily: $16−85%
Contents · per 1k pages
$0.20
Tavily: $1.60–3.20−88–94%
How the parameters map
Each Tavily parameter and what it becomes on typesearch. You rewrite one request; your agent’s logic stays.
| Tavily parameter | Status | On typesearch |
|---|---|---|
| query | Same | Same. Or up to five queries, judged together. |
| max_results | Same | Same, up to 50. |
| search_depth | Mapped | ultra-fast → mode: "ultra" · fast → "fast" · basic → "normal" · advanced → "deep" |
| time_range, start_date, end_date | Mapped | day, week, month, year → days: 1, 7, 30, 365 · start_date, end_date → published_after, published_before |
| include_domains, exclude_domains | Same | Same, up to 20 each. A domain includes its subdomains. |
| topic | Limited | We index news and editorial content: "news" is what you get. finance → sections. |
| include_answer | Limited | We don’t write answers: essential returns the key verbatim excerpts across results. |
| include_raw_content, chunks_per_source | Limited | highlights: verbatim excerpts. Never the full page. |
| country, language | Not supported | No country filter: use include_domains or sources for regional outlets. |
| include_images | Not supported | No images in results. |
| /extract | Mapped | /v1/contents: excerpts from up to 10 URLs per request, not the full page. |
What you gain
- A calibrated relevance probability on every result
- Top results read and verified before they’re ranked
- Typed questions answered per result, in the same call
- Tone, same-story dedupe and who-published-first
- Up to five queries per request
- Hard cost caps per request
When Tavily is still the better fit
We’d rather you stay than churn in a month. Keep Tavily for these:
- You search the general web more than news: product pages, docs, forums.
- You need full page text (raw_content) or images with the results.
- You want the API to write the answer, or to crawl and map whole sites.
Migration
The whole migration
Before and after: the request, the key header and the fields you read. Everything else in your code stays.
-from tavily import TavilyClient+import requests-tavily = TavilyClient(api_key=TAVILY_API_KEY)-r = tavily.search("lithium royalties in Chile", topic="news", time_range="week", max_results=10)+r = requests.post("https://api.typesearch.ai/v1/search",+ headers={"Authorization": f"Bearer {TYPESEARCH_API_KEY}"},+ json={"query": "lithium royalties in Chile", "days": 7, "max_results": 10},+).json()for hit in r["results"]:print(hit["title"], hit["url"], hit["score"])
Switching from Tavily
What is the best Tavily alternative for AI agents?
If your agent mostly searches news, typesearch does it 5.7× cheaper: $1.40 per 1,000 searches, against $8 for a basic Tavily search. Instead of a similarity score, every result carries the probability that it is about your query, and the top ones are read and verified.
Is there a drop-in replacement for the Tavily SDK?
Not yet: typesearch is drop-in for the Exa, Tako and Perplexity SDKs. From Tavily you rewrite one request — the parameters map one to one, see the table above — and keep the rest of your agent.
How much cheaper is typesearch than Tavily?
Search: $1.40 against $8. Deep search: $2.40 against $16. Contents · per 1k pages: $0.20 against $1.60–3.20. Tavily charges $0.008 per credit: 1 credit per basic search, 2 per advanced, and 1 per 5 extracted pages.
When should I stay with Tavily?
If you search the general web more than news, or you need full page text, images, crawling, or answers written by the API.
Switch from Tavily
Free key, one request to rewrite, 5.7× cheaper.