Typesense
Open-source, in-memory search engine with typo tolerance, vector search and a managed Typesense Cloud.
- Price
- Not public on the page; priced by cluster size in the calculator
- Free plan
- No
- Setup
- Needs a developer
- Up and running
- A day for the first version on Cloud; a day or two more to run your own server with backups.
What Typesense is for
Typesense is a search engine for your app's search box, with typo tolerance, filters and search by meaning (vector search). It's open source: run it on your own server for free, or rent a dedicated cluster on Typesense Cloud. On Cloud you pay by the hour for the server size, not per search, so a traffic spike doesn't blow up the bill.
How it plugs into your product
- 1Size the memory first: Typesense keeps the whole index in RAM, about 2 to 3 times the size of the fields you search.
- 2Create a cluster on Typesense Cloud (or install on your server) and add a card right away, since billing is weekly.
- 3Send your products or content from your backend and keep them in sync when they change.
- 4Connect the search box in your front end with a search-only key.
Does it fit your idea?
It fits if
- A large catalog or marketplace with spiky traffic, like Black Friday, where per-search pricing would scare you.
- You want keyword search and search by meaning in the same engine, without a separate vector database.
- You have a VPS and someone to look after it, and want search with no license cost.
Look elsewhere if
- You want a free plan forever on the cloud: the free allowance is 720 hours once, then the cluster bills by the hour.
- You need to know the exact price before signing up: Cloud prices only show in the configuration calculator.
- You have very little traffic and a big dataset: you pay for RAM all month even with almost no searches; Algolia's free plan may be enough.
What people only find out after building
Read this before you build on it. Each item links to the official page that says so.
The Cloud free tier is 720 hours and 10 GB of bandwidth once for the lifetime of the account, only on some configurations. It doesn't renew; after that, running clusters start billing.
Official source: cloud-help.typesense.orgThe hourly cluster fee is charged whether or not the cluster gets traffic, because the hardware is dedicated. A forgotten test cluster keeps billing.
Official source: cloud-help.typesense.orgThe whole index lives in RAM: plan 2 to 3 times the size of your searchable fields, plus 7 bytes per dimension per record for vectors. More data means a bigger, pricier cluster.
Official source: typesense.orgThe server is GPL-3.0. Running it as a separate service is fine, but if you distribute a modified version you must share the changes.
Official source: github.com
Price
Not public on the page; priced by cluster size in the calculator
One-time 720 cluster hours and 10GB of bandwidth on Cloud; self-hosting is free.
- Free tier
- 720 hours + 10GB once; free if self-hosted
- First paid plan
- Not public (per-cluster calculator)
- Self-host
- Yes, open source
Price not public · official pricing
What it costs in practice
Estimate: a month has about 730 hours, so the one-time 720 free hours cover roughly your first month on an eligible cluster. After that you pay the cluster's hourly fee all month (see the calculator; it isn't published on the page) plus bandwidth at about 9 to 12 cents per GB. 30K searches a month returning about 10 KB each is 0.3 GB, a few cents.
Estimate from the official price. Check it before you decide.
At a glance
- Free plan: no
- Open source: yes
“Not stated” means the official site doesn’t say.
Faundr
Picked your tools? Build it with AI, safely.
Faundr keeps your project’s memory, blocks what can’t be undone and tells you if it’s safe to launch, inside Claude Code and Codex.