Meilisearch Cloud alternatives: the same engine on a flat plan

Short version: Meilisearch Cloud is the company's own hosted version of an open-source engine, priced from $20 a month with no free plan. The usage plan is $30 for 100K documents and 50K searches; the smallest instance is about $23 for half a vCPU and 1 GB of RAM. Those are reasonable prices for a dedicated search box. They are also a second bill for a second vendor when the rest of your stack already lives somewhere. Layerbase runs Meilisearch 1.52 on the Pro plan, and Pro is $15/month, next to the Postgres and the Valkey the search index is built from. If your index is small enough for a shared pool, that is the whole argument. If it is not, I will tell you where the pool ends.

A note on who this post is for. We already have Algolia alternatives and a migration guide from Algolia to Meilisearch, both of which send you toward Meilisearch. This one is for people already on Meilisearch Cloud who are asking whether the Cloud, specifically, is the part they need.

The price ladder

Verified 2026-09-06 from meilisearch.com/pricing:

There is a 14-day trial with no card. There is no free plan, and there has not been one for a while. Self-hosting is free in the sense that the binary is open source, and the pricing page says so, but that is a server you run.

Meilisearch's own blog owns the search results for every "X alternatives" query in this category, and it is well done. What none of those posts includes is a flat multi-engine plan where Meilisearch is one engine among several, because Meilisearch does not sell that and nobody in the search category does. That is the option this post exists to describe.

The second-vendor problem

A search index is derived data. The documents in it came from somewhere, almost always a Postgres or a MySQL, and the thing that keeps the index fresh is a job that reads from the database and writes to Meilisearch. That means a Meilisearch Cloud customer with a hosted database is running two managed services from two vendors, with two dashboards, two invoices, two API keys in the environment, and two status pages to check when search returns nothing.

At $23 or $30 a month, the money is not the main cost. The main cost is that the search index is the piece of the stack most likely to be treated as optional, which means it is the piece most likely to be left on the trial, half-configured, or pointed at a stale instance. Putting it on the same account as the database it indexes is the boring fix.

What Layerbase actually gives you

Meilisearch 1.52. A new Meilisearch on Layerbase Cloud runs 1.52, the current line. Databases already on 1.33 stay on 1.33 until you decide otherwise; we do not upgrade a search engine underneath you.

Flat pricing. Meilisearch is a Pro-tier engine here. Pro is $15/month for up to 10 databases, 25 GB, every one of the 18 engines on Layerbase Cloud, 10 branches per database, 30-day rolling backups, and a 7-day trial. Your Postgres and your Valkey are two of those ten databases. Solo is $5/month if all you need is the Postgres and a second small database, but Meilisearch is not on Solo, so the search index is the thing that puts you on Pro.

The same API. An HTTPS endpoint with a master key, exactly the shape Meilisearch Cloud gives you. meilisearch-js, meilisearch-python, the Go and Rust SDKs, and the instant-meilisearch front-end adapter for InstantSearch all work by changing the host and the key. You generate scoped search keys the same way, with the same /keys endpoint.

Branching. A Meilisearch database branches like a Postgres does here: a copy-on-write clone of the index for a preview environment or a settings experiment, ready in seconds, deleted when the PR closes. Meilisearch Cloud has no equivalent; you create a second project and re-index.

Growing past the pool. The Pro plan's always-on pool is 1.5 GB, shared with the other databases you pin. When Meilisearch needs more, add $10/month pool blocks for +1 GB RAM, +1 vCPU and +25 GB of storage each, or take a Dedicated server from $35 a month (4 GB) for a search index that wants a box to itself.

Being straight about where the pool ends

Meilisearch keeps its index on disk and memory-maps it, so it is more forgiving of a small RAM allotment than an in-memory engine would be. Still, the resident set grows with the index, and the Pro pool is shared.

What fits comfortably: an index of tens of thousands to the low hundreds of thousands of documents, product catalogs, documentation sites, a SaaS app's records, with the query rates a normal application generates. That is most of what Meilisearch Cloud's usage plan sells, and it runs fine in the pool with room left for the Postgres beside it.

What does not fit: a 5M-document index with wide documents and several filterable attributes. That wants a few gigabytes of RAM for indexing alone, and it is not a fit on the base pool without pool blocks, and probably not without a Dedicated server. Meilisearch Cloud's larger instances sell exactly that, with dedicated CPU, and for an index that size their price is buying real hardware.

The other things the Cloud has that we do not: the analytics and monitoring dashboards with search-term and no-results reports, the managed embedders for their AI-powered hybrid search (you can still configure an embedder against your own OpenAI or Hugging Face key on ours, but nothing is managed for you), multi-region deployment, and an SLA on the Enterprise plan.

Moving an index

Meilisearch makes this a two-part copy: settings first, documents second. The master key on each side authorizes everything below.

1. Create the destination. Make a Meilisearch on Layerbase Cloud and copy the endpoint and master key from Quick Connect.

2. Copy the index settings. Settings are what make search behave the way you tuned it: searchable and filterable attributes, ranking rules, synonyms, stop words, typo tolerance. They are one JSON document per index.

SRC='https://ms-abc123.sfo.meilisearch.io'
SRC_KEY='<cloud master key>'
DST='https://<host>.cloud.layerbase.dev'
DST_KEY='<layerbase master key>'
IDX='products'

curl -s "$SRC/indexes/$IDX/settings" -H "Authorization: Bearer $SRC_KEY" > settings.json

curl -s -X PATCH "$DST/indexes/$IDX/settings" \
  -H "Authorization: Bearer $DST_KEY" \
  -H 'Content-Type: application/json' \
  --data-binary @settings.json

The PATCH creates the index if it does not exist. Check the primary key on the source with GET /indexes/products and pass it when you create the index if your documents do not make it obvious.

3. Copy the documents. Page through the source with limit and offset and post each page to the destination. Batches of a few thousand keep the request bodies sane.

LIMIT=5000
OFFSET=0
while :; do
  page=$(curl -s "$SRC/indexes/$IDX/documents?limit=$LIMIT&offset=$OFFSET" \
    -H "Authorization: Bearer $SRC_KEY" | jq -c '.results')
  [ "$page" = "[]" ] && break
  curl -s -X POST "$DST/indexes/$IDX/documents" \
    -H "Authorization: Bearer $DST_KEY" \
    -H 'Content-Type: application/json' \
    --data-binary "$page" > /dev/null
  OFFSET=$((OFFSET + LIMIT))
done

Honestly, though, the better source for the documents is your database. The index is derived from it, and re-running your indexing job against the new endpoint gives you a clean index with none of the deleted-document ghosts a long-lived instance accumulates. Use the API copy if the job is slow or if the documents have drifted from what the database holds.

4. Wait for the tasks, then verify. Document additions are asynchronous. GET /tasks?statuses=enqueued,processing on the destination until it is empty, then compare GET /indexes/products/stats on both sides for numberOfDocuments, and run a few searches you know the answers to.

5. Swap the keys. Change MEILI_HOST and MEILI_MASTER_KEY (or whatever your app names them) in your environment, regenerate any front-end search keys against the new instance, and redeploy. Keep the Cloud project alive until the search-term dashboard there goes quiet.

Where Meilisearch Cloud still wins

Stay if any of these is true:

• Your index is large. Millions of documents with rich settings want dedicated CPU and several gigabytes of RAM, which the Cloud's larger instances sell and our shared pool does not.
• You use the analytics dashboard. Search-term reports, no-result tracking, and the monitoring views are Cloud-only.
• You rely on managed embedders. The AI-powered hybrid search with vendor-managed embedding is a Cloud feature. You can configure an embedder against your own key on any Meilisearch, but nothing is managed for you here.
• You need multi-region or a contractual SLA, both of which are Enterprise-plan features there and not sold on our shared plans.
• Search is the product. If Meilisearch is the core of what you sell, the company that builds it is the right host.

The case for moving is the common one: the index is modest, the documents come from a database you already host, and you would rather the search engine be one more entry in the same dashboard than a second vendor with a second bill.

FAQ

How much does Meilisearch Cloud cost?

As of 2026-09-06, the usage plan is $30 a month for 100K documents and 50K searches, the smallest resource-based instance (0.5 vCPU, 1 GB RAM) is about $23 a month including disk, and the pricing page advertises "starting at $20/month." Enterprise is on quote.

Does Meilisearch Cloud have a free tier?

No. There is a 14-day trial with no card, and after that the smallest paid plan. Self-hosting the open-source binary is free, but that is a server you operate. Meilisearch is not free on Layerbase either; it is a Pro-tier engine, and the Pro plan is a flat $15 that also covers nine other databases.

Can I use the official Meilisearch SDKs and InstantSearch with Layerbase?

Yes. A Meilisearch on Layerbase exposes the standard HTTPS API with a master key, so meilisearch-js, meilisearch-python, the other official SDKs, and instant-meilisearch connect by changing the host and the key. Scoped search keys are generated through the same /keys endpoint.

How do I export an index from Meilisearch Cloud?

Pull the settings with GET /indexes/:uid/settings and the documents with GET /indexes/:uid/documents paged by limit and offset, then PATCH the settings and POST the documents to the new instance. If your documents come from a database, re-running your indexing job against the new endpoint is cleaner.

How big an index fits on the Layerbase Pro plan?

Comfortably: tens of thousands to the low hundreds of thousands of documents in a 1.5 GB always-on pool shared with your other databases. Beyond that, add $10/month pool blocks or move to a Dedicated server from $35 a month. A multi-million-document index with heavy settings is a case for the Cloud's larger instances.

Which Meilisearch version does Layerbase run?

New databases run 1.52. Databases created on 1.33 stay on 1.33 until you choose to move; we never upgrade an engine version underneath a running database.

The wrap-up

Meilisearch is open source, and Meilisearch Cloud is the company's fair price for running it with dedicated hardware, analytics, and managed AI features on top. If those are why you are there, stay. If the index is modest and the documents already live in a database you host, create a Meilisearch on Layerbase Cloud, copy the settings and the documents, and let search be one more line in the dashboard you already open.