Question 1

Which is cheaper for site search — Algolia, Typesense, or Meilisearch?

Accepted Answer

It depends on the shape of your traffic, not on which vendor is generically 'cheaper,' because the three bill on fundamentally different models. Algolia's self-serve Grow plan and Meilisearch Cloud are usage-based — you pay per 1,000 searches and per 1,000 records or documents beyond a free allowance, so the bill scales with how hard people actually use the search box. Typesense Cloud is resource-based — you provision a dedicated cluster sized by RAM/CPU and pay a fixed fee for it every month no matter how many searches you run, with no per-query charge at all. The practical result is that a site with a large catalog but modest search volume (a big docs site or content archive that isn't a top-10 destination) tends to land cheaper on Typesense's flat fee, while a site with a small catalog and either very light or very bursty query volume can land cheaper on Algolia's or Meilisearch's free allowance and low per-unit overage. There is no universal winner — run your own monthly searches and records through this calculator rather than trusting a generic 'X is cheaper' answer, because that answer is only true for one specific traffic shape.

Question 2

Why doesn't Typesense Cloud charge per search?

Accepted Answer

Typesense Cloud sells you a dedicated cluster — a fixed slice of RAM and CPU — and bills a fixed fee for every hour that cluster is running, regardless of how many searches or indexing operations you send it, per Typesense's own billing documentation. That's a fundamentally different pricing primitive from Algolia or Meilisearch's per-operation billing: instead of metering usage, Typesense meters the infrastructure you've reserved, so running one search or one million searches against the same cluster costs the same fixed monthly fee. The cluster size — and therefore the price — is really a data-sizing decision: you provision enough RAM to hold your records comfortably in memory (Typesense's own guidance is roughly 2-3x your on-disk dataset size), and the query volume determines whether that same cluster needs more CPU headroom for throughput, not a different pricing tier. The two costs that do scale independently of the flat cluster fee are outbound bandwidth, billed per GB, and optional priority support plans — everything else is pinned to the cluster you chose, which is exactly why heavy query volume against a modest record count is where Typesense tends to win the cost comparison.

Question 3

How does Algolia's Grow-plan pricing actually work, and what about Meilisearch Cloud?

Accepted Answer

Algolia's self-serve Grow plan includes 10,000 search requests and 100,000 records per month at no base cost, then bills $0.50 per additional 1,000 search requests and $0.40 per additional 1,000 records beyond those allowances — a pure pay-as-you-go model with no monthly minimum, confirmed on Algolia's own pricing and billing-support pages. A separate Grow Plus tier keeps the same free allowances but charges $1.75 per additional 1,000 searches in exchange for AI-ranking and personalization features most comparison shoppers don't need. Meilisearch Cloud instead sells named plans: Build at $30/month includes 50,000 searches and 100,000 documents, with overage at $0.40 per 1,000 extra searches and $0.30 per 1,000 extra documents; Pro at $300/month raises the included allowance to 250,000 searches and 1,000,000 documents with cheaper overage rates. Meilisearch also offers a separate resource-based cluster tier in beta, similar in spirit to Typesense's model, for teams that want to pin down a flat fee instead. Both Algolia and Meilisearch's headline plans are usage-metered; only Typesense's core offering is fully flat regardless of query volume.

Question 4

How is this different from the AI Search / Answer Engine Cost Calculator and the Vector DB Storage Cost Calculator?

Accepted Answer

All three tools price something with 'search' in the name, but they're pricing three unrelated products. The AI Search / Answer Engine Cost Calculator prices a Perplexity-style pipeline where an LLM calls a live web-search API on every query and synthesizes a cited answer from the results — that's about answering open-ended questions against the whole internet, with an LLM token cost stacked on top of a search-API fee, and it has nothing to do with a search box on your own website. The Vector DB Storage Cost Calculator prices the raw infrastructure — a Pinecone- or Weaviate-style vector database storing embeddings — that you'd use as the retrieval backend if you were building your own semantic search or RAG pipeline from scratch, with no query interface, ranking, typo-tolerance or facets included. This calculator prices a third, distinct category: a drop-in, hosted 'instant search as you type' widget — typo-tolerant keyword and facet search with a ready-made UI, the kind that powers e-commerce catalogs, documentation sites and content search boxes — billed by the vendor per query and/or per record rather than something you assemble yourself from vector primitives. If you're wiring an LLM to live web search, use the answer-engine calculator; if you're building custom vector retrieval, use the Vector DB calculator; if you're adding a search box to a site or app via Algolia, Typesense or Meilisearch, this is the right tool.

サイト検索 API コスト計算ツール

コストの内訳

クロスオーバー: Typesense の定額料金が Algolia の検索ごとの請求を上回るのはいつですか?

レコード数に応じた検索ボリューム別の費用

使用量従量制とリソースフラット — 実際に重要な決定

この計算機の仕組み