Resource-based pricing
Resource-based projects are billed at an hourly rate based on the resource tier you select (Memory and vCPU). You pay for the resources you provision, regardless of how many searches you run or documents you index. The Cloud UI shows the hourly rate and estimated monthly cost for each tier at project creation and in the project settings. Pricing may vary slightly by region.Usage-based pricing
Usage-based projects are billed on what your project actually consumes. There are three plans:| Plan | Included searches | Extra searches | Included documents | Extra documents | Resources |
|---|---|---|---|---|---|
| Build | 50K/month | $0.40 per 1,000 | 100K | $0.30 per 1,000 | Shared |
| Pro | 250K/month | $0.30 per 1,000 | 1M | $0.20 per 1,000 | Dedicated |
How usage-based billing is charged
- Plan cost: the base plan fee (Build or Pro) is charged upfront at the start of each billing cycle.
- Extra usage: searches and documents beyond the included quota are charged at the end of the billing cycle, once the total is known.
- Cancellation: if you cancel your plan before the end of the month, the unused portion of the base plan fee is prorated and returned as a credit.
- Outstanding usage: if you remove your payment method while extra usage charges are still outstanding, Meilisearch will follow up to collect the owed amount.
Shared billing rules
Regardless of billing model:- Per-project billing. Each project is billed independently. A team with multiple projects is billed the sum of all project charges.
- Prorated daily cycle. Creating or deleting a project mid-day adjusts the charge proportionally.
- No per-seat fees. Adding team members does not affect billing.
Regions and pricing
Pricing may vary slightly by region. The Cloud UI shows the exact price for your selected region.| Region | Location |
|---|---|
FRA | Frankfurt |
LON | London |
SGP | Singapore |
JPN | Japan |
SFO | San Francisco |
NYC | New York |
Choosing a resource tier
For resource-based projects, the most important factor is RAM: Meilisearch keeps indexes in memory for fast search, so your instance needs enough RAM to hold your index comfortably.Step 1: Estimate your index size
Your index size depends on how many documents you have and how large each document is. Use these typical document sizes as a starting point:| Document type | Avg size | Examples |
|---|---|---|
| Small | ~1 KB | SaaS records, simple product listings with few filters |
| Medium | ~3 KB | E-commerce products with descriptions and ~10 filterable attributes |
| Large | ~8 KB | Articles, blog posts, rich content |
| AI (with vectors) | ~12 KB | Any document type with vector embeddings |
| Documents | Avg size | Estimated index size |
|---|---|---|
| 100K | 3 KB (medium) | ~1.4 GB |
| 500K | 3 KB (medium) | ~7 GB |
| 100K | 12 KB (AI) | ~5.7 GB |
| 1M | 8 KB (large) | ~37 GB |
Step 2: Choose a tier with enough RAM
Choose the smallest tier where RAM exceeds your estimated index size. Leave headroom for query cache and peak usage.| Tier | vCPU | RAM | Suitable for |
|---|---|---|---|
| XS | 0.5 | 1 GB | Development and testing |
| S | 1 | 2 GB | Up to ~80K small documents |
| M | 2 | 4 GB | Up to ~80K medium or ~160K small documents |
| L | 2 | 8 GB | Up to ~400K medium documents |
| XL | 4 | 16 GB | Up to ~800K medium or ~300K AI documents |
| 2XL | 8 | 32 GB | Up to ~1.6M medium or ~600K AI documents |
| 4XL | 16 | 64 GB | Up to ~3M medium or ~1.2M AI documents |