MongoDB Atlas Vector Search API pricing
Vector search native to Atlas, letting teams add embeddings to an existing operational DB with no new service.
Free tier or trialBilled atlas cluster consumption-based·Verified Jun 27, 2026
Free tier
M0 cluster: 512MB storage, shared
Plans and prices
| Plan | Price | What it covers |
|---|---|---|
| Free (M0) | $0/mo | 512 MB storage, shared RAM/vCPU; for learning and prototyping. Vector Search available on shared tiers with limits. |
| Flex | $8-$30/mo | Usage-based ($0.011-$0.041/hr), 5 GB storage, tiered by ops/sec; for dev/test and unpredictable traffic. |
| Dedicated (M10+) | from $0.08/hr (~$57/mo) | M10 = 2 vCPU / 2 GB RAM; scales to M300 (96 vCPU / 384 GB) at $21.85/hr. Required for production with 99.995% SLA. |
| Vector Search Nodes (High-CPU) | $0.12-$3.26/hr per node | Dedicated search compute, billed separately, two-node minimum. S20 ($0.12) to S80 (64 vCPU / 128 GB, $3.26). |
| Vector Search Nodes (Storage-optimized) | ~$1.04/hr for s50 (2 nodes) | s50 = 32 GB RAM / 4 vCPU used in MongoDB benchmarks for scalar-quantized 15.3M-vector workloads. |
What to watch for on the bill
- Pricing is the most common complaint across G2/PeerSpot - expensive and hard to predict versus self-hosting
- HNSW indexes need ~1.2-1.5x raw vector size in RAM and must fit in memory, making memory planning critical and upgrades costly
- Production vector workloads require separately-billed Search Nodes with a two-node minimum, adding base cost
The pricing page, as we saw it
Source: https://www.mongodb.com/pricing. Prices change. Confirm on the vendor page before you commit.
Compare MongoDB Atlas Vector Search
- Pinecone vs MongoDB Atlas Vector Search
- MongoDB Atlas Vector Search vs Zilliz Cloud (Milvus)
- MongoDB Atlas Vector Search vs Supabase Vector (pgvector)
Other Vector Database APIs
Pinecone · Zilliz Cloud (Milvus) · Supabase Vector (pgvector) · Qdrant Cloud · Weaviate Cloud · Turbopuffer