Hybrid search
Pure keyword search misses paraphrases; pure semantic search sometimes misses an exact part number or a product code it should have nailed instantly. This endpoint runs both in the same call and fuses the two rankings into one result set, so a query gets the benefit of exact-term precision and conceptual understanding at once.
Run it online
Run this on our servers with your account. Free tools run in your browser; this one bills your KIT balance per the price above.
The tradeoff neither approach solves alone
Keyword search is precise on exact terms — a SKU, a model number, a legal citation — but blind to paraphrase. Semantic search understands paraphrase but can occasionally rank a loosely related concept above a document that contains the literal term the user typed, particularly when that term is rare. Neither weakness is fatal alone, but a search box that only does one will systematically underperform on some share of queries, and you rarely know in advance which type a given search will be.
What the fusion actually does
POST a query and your candidate set to /search/hybrid and the task runs a keyword-based match and a meaning-based match in parallel, then combines both rankings into a single fused score per document rather than making you pick a winner or run two separate calls and merge results yourself. A document that scores well on both signals rises to the top; one that only scores well on one still surfaces, just lower, instead of vanishing entirely the way it would in a single-method system.
A well-established idea in information retrieval
Combining lexical and vector signals isn't new — search engineers have used rank fusion techniques for years to blend independently scored result lists into one ordering that outperforms either alone. What's changed is that meaning-based ranking used to require specialized infrastructure most teams couldn't justify building, while keyword search has been a commodity for decades; this endpoint packages the fusion of both as a single call so you don't run two systems and stitch the outputs together yourself.
Where hybrid outperforms either method alone
E-commerce search is the clearest case: a shopper might type an exact model number one minute and 'waterproof shoes for hiking' the next, and both need to work well from the same search box. Documentation search, support ticket lookup, and enterprise search over mixed content all benefit the same way, since real queries are a mix of exact terms and loose descriptions and rarely announce in advance which one they are.
Pricing and delivery
Billing is a flat rate per request plus a per-query charge, the same structure whether the query leans keyword-heavy or fully conceptual, with no separate charge for running both signals. The call is asynchronous: you get a task_id immediately and the fused, ranked result set arrives by signed webhook or a signed link valid for 24 hours; failed tasks retry automatically up to three times and are never billed if they still can't complete.
What you can do with it
E-commerce search box
A retailer's search handles both 'SKU-88213' and 'lightweight running jacket' from the same input field, without needing separate logic for each query style.
Technical documentation search
A developer portal matches exact function names and error codes as well as loosely described problems like 'why does my request time out.'
Support ticket deduplication
A helpdesk surfaces prior tickets that mention the same product code or describe the same issue in different words, cutting duplicate investigation work.
Enterprise search over mixed content
A company search tool ranks contract numbers, policy IDs, and plain-language questions correctly in the same index without maintaining two separate search systems.
FAQ
What is hybrid search and how is it different from semantic search alone?
Hybrid search combines keyword matching with meaning-based ranking into a single fused result set, catching both exact terms and paraphrased queries that pure semantic search can occasionally rank lower than it should.
Do I need to configure the weighting between keyword and vector signals?
No, you send the query and candidates to the endpoint and it handles the fusion; there's no separate infrastructure or tuning step required on your side.
Is hybrid search free to use?
There's no free tier — free tiers get abused and slow everyone down. Access runs on a prepaid ForHosting KIT balance: top up from $10.00 (it never expires) and each request is charged at its published price, so a call with no balance returns HTTP 402. No subscription, no tokens, no invented credits, and a failed task is never charged.
How do I get the results back?
The endpoint returns a task_id immediately and the fused ranked results arrive by signed webhook or a signed link valid for 24 hours.
When should I use hybrid search instead of pure semantic search?
Use hybrid whenever queries might include exact terms like SKUs, codes, or names alongside natural-language descriptions, which is the case for most real-world search boxes.
Can hybrid search handle large product catalogs or document sets?
Yes, you submit each query as its own async request against your candidate set with no artificial limit on request volume, billed per the standard per-request and per-query rate.
Does hybrid search cost more than semantic search alone?
No, pricing is the same base plus per-query rate; running both signals in one call is included, not billed as two separate operations.
What happens if a hybrid search request fails?
The task retries automatically up to three times, returns a clear error if it still can't complete, and a failed request is never charged.
For developers — API access
Everything on this page is available programmatically. This section is for teams who want to wire it into their own systems; everyone else can just use the tool above.
API endpoint
Prefer to automate it? One authenticated POST creates the task; the result comes back by webhook or a signed link. The same capability also runs here on the web, and soon from our app, email and Telegram.
Call it from your stack
curl -X POST https://api.kit.forhosting.com/search/hybrid \
-H "Authorization: Bearer $KIT_KEY" \
-H "Content-Type: application/json" \
-d '{"input":"…"}'const res = await fetch("https://api.kit.forhosting.com/search/hybrid", {
method: "POST",
headers: {
"Authorization": `Bearer ${process.env.KIT_KEY}`,
"Content-Type": "application/json"
},
body: JSON.stringify({
"input": "…"
})
});
const { task_id } = await res.json();import os, requests
res = requests.post(
"https://api.kit.forhosting.com/search/hybrid",
headers={"Authorization": f"Bearer {os.environ['KIT_KEY']}"},
json={
"input": "…"
},
)
task_id = res.json()["task_id"]<?php
$res = file_get_contents("https://api.kit.forhosting.com/search/hybrid", false, stream_context_create([
"http" => [
"method" => "POST",
"header" => "Authorization: Bearer " . getenv("KIT_KEY") . "\r\nContent-Type: application/json",
"content" => '{"input":"…"}',
],
]));
$task = json_decode($res, true);body := bytes.NewBufferString(`{"input":"…"}`)
req, _ := http.NewRequest("POST", "https://api.kit.forhosting.com/search/hybrid", body)
req.Header.Set("Authorization", "Bearer "+os.Getenv("KIT_KEY"))
req.Header.Set("Content-Type", "application/json")
res, _ := http.DefaultClient.Do(req)Example request
{
"input": "…"
}Example response
{
"task_id": "tsk_a1b2c3d4e5f6a1b2c3d4e5f6",
"type": "search.hybrid",
"status": "queued",
"_links": {
"result": "/tasks/tsk_…/result"
}
}The API is asynchronous: the call returns a task_id immediately and the result arrives by webhook. Polling is capped at 1 req/s per task.
Pricing
Published price — no tokens, no invented credits. A failed task is never charged.
Limits
max_chunks | 10000 |
max_tokens | 20000 |
Errors
| HTTP | Code | Meaning |
|---|---|---|
401 | unauthorized | Missing or invalid API key. |
402 | insufficient_balance | Your balance doesn't cover the task price. |
404 | unknown_type | That task type doesn't exist. |
429 | rate_limited | Too many requests. Use the webhook instead of polling. |