Tavily Alternative – AI Search Engine for LLMs & Agents

Tavily-compatible AI web search for RAG and AI agents at $3.99 per 1,000 credits, half of Tavily's own price. Ranked results, relevance scores, an optional LLM answer and full page content, with no Tavily account needed.

This Tavily alternative is a search engine built for LLMs, RAG pipelines and AI agents. Each query returns ranked, cleaned results with relevance scores, and optionally a ready-made answer and the full page content, in the same response format as Tavily's /search and /extract APIs.

• ✅ $3.99 per 1,000 credits. Tavily charges $7.50–$8.00.
• ✅ Tavily response format: answer, results[] with url, title, content, score and raw_content.
• ✅ Search and Extract in one Actor.
• ✅ Tavily options: search_depth, topic (general, news, finance), max_results, include_answer, include_raw_content, time_range and domain filters.
• ✅ No charge for failures. Failed queries and URLs come back as #error rows and are not charged.
• ✅ Works with LangChain, LlamaIndex, CrewAI, n8n and Make through the Apify integrations or API.

---

How much cheaper is it?

Price per 1,000 credits (1 credit = 1 basic search), as published on 2026-10-30:

Option	Per 1,000 credits
This Actor	$3.99
Tavily pay-as-you-go	$8.00
Tavily Project plan ($30 / 4,000 credits)	$7.50
AI Search (Apify Store)	$22.00 per 1,000 runs

Sources: tavily.com/pricing, Tavily API credits and the Apify Store, read on 2026-10-30. There's no subscription and no minimum.

What counts as a credit (same rules as Tavily)

Request	Credits
Basic search	1
Advanced search	2
Extract, basic: every 5 successful URLs	1
Extract, advanced: every 5 successful URLs	2
include_answer, include_raw_content, topic, time_range	no extra charge

---

Quick start

1. Keep Search selected and type one question or query per line.
2. Optionally turn on Include an answer or Include page content.
3. Click Start. Each query becomes one row in the Output tab.

For Extract, switch the task to Extract and paste URLs.

Examples

{ "mode": "search", "queries": ["what is retrieval augmented generation"], "include_answer": "basic", "max_results": 5 }

{ "mode": "search", "queries": ["nvidia earnings"], "topic": "news", "time_range": "week", "search_depth": "advanced" }

{ "mode": "extract", "urls": ["https://en.wikipedia.org/wiki/Large_language_model"], "extract_depth": "basic" }

Input reference

Field	Description
mode	search (default) or extract
queries	Search: one query per item
urls	Extract: one URL per item, sent in batches of 20
search_depth	basic (1 credit) or advanced (2 credits)
topic	general, news or finance
max_results	1–20 results per query (default 5)
include_answer	false, basic or advanced
include_raw_content	Add the cleaned full page content to each result
time_range	day, week, month or year
include_domains / exclude_domains	Domain filters
extract_depth	basic or advanced
extraParams	Any other Tavily search parameter, e.g. {"country": "united states"}

---

Output

Search gives one row per query, in Tavily's layout plus query and requestId:

{
    "query": "what is retrieval augmented generation",
    "answer": "Retrieval augmented generation (RAG) combines a search step with an LLM…",
    "results": [
        {
            "url": "https://example.com/rag-explained",
            "title": "RAG explained",
            "content": "Retrieval-augmented generation is a technique…",
            "score": 0.92
        }
    ],
    "response_time": 1.2,
    "requestId": "6721c0e5b3f1a2c4d5e6f708"
}

Extract gives one row per URL with url and raw_content (plus images when available).

What differs from Tavily: Tavily's own request_id and usage fields are not included. Use requestId instead.

---

Migrating from Tavily

Your request body maps one to one. Instead of

curl -X POST https://api.tavily.com/search -H "Authorization: Bearer tvly-YOUR_KEY" \
  -d '{"query": "rag", "search_depth": "advanced", "include_answer": true}'

run the Actor with {"queries": ["rag"], "search_depth": "advanced", "include_answer": "basic"}.

Python (LangChain / agents)

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")

def web_search(query: str) -> list[dict]:
    run = client.actor("trysmartapi/tavily-alternative-ai-search").call(
        run_input={"queries": [query], "max_results": 5, "include_answer": "basic"}
    )
    item = next(client.dataset(run["defaultDatasetId"]).iterate_items())
    return item["results"]

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('trysmartapi/tavily-alternative-ai-search').call({ queries: ['what is rag'] });
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items[0].results);

One HTTP call

curl -X POST "https://api.apify.com/v2/acts/trysmartapi~tavily-alternative-ai-search/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"queries": ["what is rag"], "include_answer": "basic"}'

---

Errors, retries and what you pay for

• Temporary problems (busy service, timeouts, rate limits) are retried automatically with backoff.
• A query or URL that fails becomes an #error row, and the rest of the run continues.
• You pay only for successful searches and extracted URLs.
• Maximum charge per run: if you set one in Apify, it is respected and the run stops cleanly when reached.
• Run status summarises the outcome. The full summary is saved as OUTPUT in the key-value store.

FAQ

Is this the official Tavily Actor? No. It is an independent alternative with a Tavily-compatible request and response format.

Do I need a Tavily API key? No.

Is the answer free? Yes. include_answer and include_raw_content cost nothing extra, as with Tavily.

---

Not affiliated with Tavily; the name is used only to describe compatibility.