MongoDB Atlas Pricing Explained (October 2026)

MongoDB Atlas is billed hourly and invoiced monthly. As of October 2026, MongoDB's pricing page lists three Atlas Core tiers: Free at $0, Flex at $0.011 per hour capped at $30 a month, and Dedicated clusters from $0.08 per hour, which the page puts at $56.94 a month for the smallest M10 configuration. Storage beyond the default, backups, data transfer and add-ons are charged on top.

All figures below are list prices taken from mongodb.com/pricing and the Atlas billing documentation on 11 October 2026. Prices vary by cloud and region; check the Atlas pricing calculator for your own configuration.

Dedicated tiers are priced per hour and differ by cloud provider and region. The pricing page's dedicated table shows these base prices for AWS us-east-1, including each tier's default storage:

Three details change the real number. Atlas charges the cluster and storage cost for every data-bearing node, so extra electable, read-only or analytics nodes and every additional shard multiply it, and sharded clusters add config servers at a separate rate. Multi-region clusters cost more than the preferred-region price shown at creation. And M10 and M20 run on burstable instances that can be CPU-throttled under sustained load, which is why MongoDB recommends M30 and above for performance-critical workloads.

Atlas Infinite, which separates compute from storage, is in public preview from $0.09 per hour.

• Storage. Default storage is in the hourly price. If you customise capacity, Atlas bills the full amount, not just the increase. Faster IOPS costs extra.
• Cloud Backup. Snapshot storage is billed per GB per month: $0.14 to $0.19 on AWS, $0.08 to $0.12 on Google Cloud and $0.34 to $0.65 on Azure. Continuous cloud backup, which adds point-in-time restore, is billed on snapshots plus the stored oplog in tiers: on AWS $1.00 to $1.55 per GB between 5 and 100 GB, falling to $0.25 to $0.40 above 500 GB. Copies in other regions roughly double storage.
• Data transfer. Inbound is free. On AWS, same-region transfer is $0.00 to $0.01 per GB, cross-region $0.02 to $0.15 and out to the internet $0.09 to $0.16. Free and Flex clusters are not charged for outbound data.
• Search and Vector Search Nodes. Dedicated search hardware is billed hourly; on AWS us-east-1 high-CPU nodes run from $0.12 per hour (S20) to $3.26 per hour (S80).
• Online Archive and Data Federation. Data Federation charges $5.00 per TB processed with a 10 MB minimum per query, plus data returned. Online Archive adds storage per GB per day.
• Networking and security. Private endpoints carry an hourly charge per region ($0.045 per hour in us-east-1 on AWS) plus capacity units. Customer-managed encryption keys or LDAP add a 15% uplift to each cluster, except on the Atlas Enterprise and Platinum plans.
• Support. Basic support is free. Developer, Pro, Enterprise and Platinum plans are paid, with severity 1 initial response goals of 8 hours, 2 hours, 30 minutes and 15 minutes. The support page does not list plan prices.

• Tier steps double. Each tier roughly doubles the hourly rate: M30 to M40 to M50 is $0.54, $1.04, then $2.00 an hour. A working set that outgrows RAM pushes the cluster up a step, often because of an indexing problem.
• Auto-scaling without a ceiling. Compute auto-scaling moves the tier up on CPU and memory pressure, and storage on Atlas Core grows when the disk reaches 90% full.
• Backups that only accumulate. Snapshots are incremental, but Atlas keeps an extra full snapshot after a tier, storage or topology change or a failover, and the cloud provider can report snapshot usage up to the full volume size. Longer restore windows mean a larger stored oplog.
• Cross-region traffic. Applications in another region and multi-region clusters add transfer charges.
• Clusters nobody owns. Development and test clusters left running at dedicated rates. Since 22 January 2026 Atlas no longer supports M2 and M5; they were migrated to Flex, so a former flat-rate cluster now bills by ops per second.

• Fix queries before buying hardware. Use the Performance Advisor's index suggestions and query projections; a smaller working set can keep you on a lower tier.
• Cap auto-scaling. Set a minimum and maximum cluster size; Atlas will not scale outside that range.
• Pause or downsize non-production clusters. M10 and larger clusters can be paused for up to 30 days, after which Atlas resumes them; only storage is billed while paused. Flex clusters cannot be paused, but at $8 to $30 a month they are the cheaper home for development.
• Tune the backup policy. The default policy takes hourly snapshots (every 6 hours, or 12 on NVMe) kept 7 days, plus daily, weekly, monthly and yearly snapshots. Shorten retention and the restore window where your recovery objectives allow, and limit cross-region copies.
• Keep traffic local. Run applications in the cluster's cloud region, enable driver network compression and return only the fields you need.
• Tier cold data. TTL indexes or Online Archive move old documents off primary storage.
• Watch the spend. Billing alerts and the Billing Cost Explorer catch a runaway cluster before the invoice does.

For larger estates, Atlas offers prepaid, monthly commitment and marketplace subscriptions. The documentation describes how they are invoiced but publishes no discount rates, so any reduction is negotiated with MongoDB.

Atlas bundles hardware, operations, backup and upgrades into one usage bill. The alternative is to run MongoDB yourself:

• Community Server. No licence fee under the SSPL. You pay for servers, storage and the engineers who run replica sets, upgrades and backups, and MongoDB, Inc. does not support it.
• Enterprise Advanced. A commercial subscription for self-managed deployments with Enterprise Server, Ops Manager, the Kubernetes Operator and vendor support. MongoDB quotes it through sales rather than publishing a price.
• Either, with third-party support. Independent providers support Community, Enterprise and Atlas deployments. Our MongoDB support options guide compares the routes.

A fair comparison includes backup storage, monitoring, on-call time and upgrade work, not only instance prices. Moving in either direction is covered in our MongoDB migration guide, and the backup choices on each side are in how to back up a MongoDB database. If the reason you are pricing Atlas is an old self-managed cluster on 4.2 to 6.0 that cannot be upgraded yet, OSSeva, AceMQ's extended support platform, ships patched builds for those versions while you decide.

Most Atlas cost problems trace back to sizing, indexing and backup policy rather than price lists. AceMQ's MongoDB support covers Atlas, Enterprise Advanced and Community deployments, including reviews of why a cluster is on the tier it is on. Our WiredTiger cache eviction case shows the kind of working-set problem that drives tier upgrades, and database support covers MongoDB alongside the rest of your data stack.

• MongoDB: Pricing (Atlas tiers, dedicated, Flex and Search Node tables, FAQ), checked 11 October 2026
• MongoDB Atlas: Cluster Configuration Costs (storage, backup, nodes)
• MongoDB Atlas: Estimate Atlas Flex Cluster Costs (M2/M5 retirement)
• MongoDB Atlas: Data Transfer Costs
• MongoDB Atlas: Additional Services (private endpoints, 15% security uplift)
• MongoDB Atlas: Data Federation Costs
• MongoDB Atlas: Billing Breakdown and Optimization
• MongoDB Atlas: Manage Subscriptions
• MongoDB Atlas: Auto-Scaling
• MongoDB Atlas: Pause, Resume, or Terminate a Cluster
• MongoDB Atlas: Configure a Backup Policy (default policy)
• MongoDB Atlas: Flex Limitations
• MongoDB: Atlas Support Plans
• OSSeva: MongoDB extended support