MongoDB Atlas Pricing: Free Tier and Cluster Cost▋

How much does MongoDB Atlas cost?

MongoDB Atlas costs $0 a month on the free M0 tier, $8 to $30 a month on Flex, and from $56.94 a month on the dedicated tier. The dedicated ladder starts at an M10 at $0.08 an hour and runs to an M700 at $33.26 an hour. Backup and data transfer are billed separately on top of every paid tier.

The number most teams quote each other is the M10 figure, because that is the first tier that gives you dedicated compute, and it is the one nearly everyone lands on when a prototype turns into a product. At $56.94 a month it looks modest. What makes the real bill larger is that Atlas prices several things that feel like part of a database but are billed as their own meters.

Three of those matter in practice. Backup storage is $0.14 per GB per month, charged on the datafile size on disk rather than the logical size of your documents. Data transfer is $0.01 per GB within a region, $0.02 per GB across regions and $0.09 per GB out to the internet. Atlas Search, if you enable it, runs on separate instances that start at $0.13 an hour, which is effectively a second cluster.

A worked example. An M30 at $387.62 a month holding 300 GB, with backups retained and 500 GB a month of responses going out to the internet, is $387.62 for compute, about $42 for backup and about $45 for egress. That is roughly $475 a month, and the cluster tier accounts for barely eight tenths of it.

Is MongoDB Atlas free?

Yes, MongoDB Atlas has a free tier called M0 that never expires and does not require a payment method. It gives you 512 MB of storage, shared RAM and vCPU, 32 MB of sort memory and up to 100 operations per second. It is a genuine free tier rather than a trial, and MongoDB does not put a time limit on it.

The constraint that ends the free tier for most people is storage rather than throughput. 512 MB sounds generous until you store documents with embedded arrays, or keep a few months of event data, at which point it fills quickly. The 100 operations per second ceiling is usually the second wall, not the first.

M0 also has feature limits beyond capacity. Atlas Charts on an M0 cluster refreshes once every four hours and reports weekly, where dedicated and Flex clusters get unlimited refreshes. Data transfer fees are not assessed on M0, M2 or M5 at all, which is worth knowing because transfer is one of the least predictable lines on a paid cluster.

What are the MongoDB Atlas free tier limits?

The M0 free tier limits are 512 MB of storage, 32 MB of sort memory, shared RAM and vCPU, and up to 100 operations per second. There is no monthly cost and no expiry date. Data transfer charges do not apply to M0.

Those are the published figures. In day to day use the sort memory limit is the one that catches people out, because it is not a number anyone thinks about until a query that sorts a large result set fails rather than running slowly. If you are hitting sort memory errors on M0, that is a signal to add an index rather than to upgrade the tier.

When M0 does run out, the instinct is to jump to an M10 at $56.94 a month. Flex sits in between at $8 to $30 for 5 GB, which is ten times the storage of M0 for a seventh of the M10 price. Skipping that step is the single most common way teams overpay for Atlas early on.

How much does an M10 cluster cost?

An M10 cluster costs $0.08 an hour, which MongoDB publishes as $56.94 a month for the AWS N. Virginia region. That buys 2 GB of RAM, 2 vCPUs and a storage range of 10 to 128 GB. It is the entry point to the dedicated tier and the first tier with compute that is not shared.

The M10 price is per cluster, per hour, and it does not stop when traffic does. A staging cluster left running all month costs the same $56.94 as a production one serving real users. Teams running separate dev, staging and production clusters are therefore paying about $171 a month before a single byte of backup or egress is added.

Above M10 the ladder is driven by RAM and each step is a large one. M20 doubles RAM to 4 GB and costs $146.72, which is 2.6 times the M10. M30 doubles again to 8 GB at $387.62. There is no gentle slope here, so it pays to check whether your working set genuinely exceeds the RAM of the tier you are on before stepping up.

MongoDB also publishes low CPU variants of the dedicated tiers at roughly a 26 percent lower hourly rate. If your workload is memory bound rather than CPU bound, which many document store workloads are, those variants are worth pricing before you accept the standard rate.

What is MongoDB Atlas Flex?

Flex is the tier between free and dedicated, priced from $0.0110 an hour, about $8 a month, up to $0.0411 an hour, about $30 a month. It includes 5 GB of storage, runs on shared RAM and vCPU, and carries no data transfer charge.

Flex is banded by operations per second rather than metered continuously. MongoDB publishes five bands: 0 to 100 operations per second at $8 a month, 100 to 200 at $15, 200 to 300 at $21, 300 to 400 at $26, and 400 to 500 at $30. Your bill lands in the band your usage reaches, so it is capped and predictable in a way a purely usage based tier is not.

The absence of a data transfer charge is the underrated part. On a dedicated cluster, egress at $0.09 per GB to the internet is the line that varies most between months and is hardest to forecast. Flex removes it entirely, so a $30 Flex bill is genuinely $30.

How much does MongoDB Atlas backup cost?

MongoDB Atlas backup storage costs $0.14 per GB per month, charged on the datafile size on disk rather than the logical size of your data. It is billed on top of the cluster rate on every paid tier.

The on-disk basis is the detail that makes this line larger than people expect. Your documents may total less than the space MongoDB actually occupies once indexes and storage overhead are counted, and the backup charge follows the larger of those numbers. A cluster carrying 500 GB on disk is $70 a month in backup alone, which on an M30 is a fifth of the compute cost again.

Retention is the lever you control. Longer retention windows and more frequent snapshots both increase stored GB. If backup is a meaningful share of your invoice, the fix is usually to shorten retention on non-production clusters rather than to change tier.

Does MongoDB Atlas charge for data transfer?

Yes. MongoDB publishes data transfer at $0.01 per GB in the same region, $0.02 per GB across regions and $0.09 per GB out to the internet. Transfer fees are not assessed on M0, M2 and M5 instances.

The internet rate is the one that produces surprise bills. It applies to data leaving Atlas for a destination outside the cloud provider's network, which includes an application server hosted somewhere else. A read heavy API returning 1 TB of responses a month to an application outside the region pays about $90 in transfer, on top of whatever the cluster costs.

The structural fix is colocation. Putting the application in the same region as the cluster drops the rate from $0.09 to $0.01 per GB, a nine fold reduction with no change to the code. Putting the database on the same server as the application removes the charge altogether, which is one reason teams with steady, predictable load often move off managed clusters entirely.

Why is my MongoDB Atlas bill so high?

The usual answer is that the cluster tier is only part of the bill. Backup at $0.14 per GB, internet egress at $0.09 per GB and Atlas Search nodes starting at $0.13 an hour are billed separately, and together they routinely add 20 to 50 percent on top of the cluster rate.

Idle clusters are the second cause. Atlas dedicated tiers bill per hour regardless of query volume, so a development cluster nobody has touched in three weeks still costs $56.94 a month at M10. Teams that spin up a cluster per branch or per engineer accumulate these quickly, and because each one is individually cheap, nobody notices until the total is large.

The third is over-provisioning on RAM. Because the tier ladder is steep, a team that moves to M30 to fix a slow query is paying $387.62 a month where an index on an M10 at $56.94 might have solved it. Checking query plans before changing tier is the highest leverage thing you can do to an Atlas bill.

If you want the full picture before committing, the same discipline applies to every managed platform. We keep first-party rate cards for Supabase pricing and Firebase pricing, both of which price databases on completely different meters to Atlas.

How do I reduce MongoDB Atlas costs?

The four levers that actually move an Atlas bill are dropping to Flex where the dataset fits, colocating the application with the cluster, shortening backup retention, and removing idle clusters. In that order, they are cheaper to apply than changing tier.

• Move to Flex if you fit in 5 GB. $8 to $30 a month against $56.94 for an M10, and no data transfer charge at all.
• Colocate the application. Same region transfer is $0.01 per GB against $0.09 to the internet, so this is a nine fold cut on one line for a configuration change.
• Shorten retention on non-production clusters. Backup is $0.14 per GB per month on the on-disk size, and staging rarely needs the retention window production does.
• Delete idle clusters. Dedicated tiers bill per hour whether or not they are queried. One forgotten M10 is $683 a year.
• Index before you upgrade. The tier ladder roughly doubles in price at each step, so a query plan fix is worth several hundred dollars a month at the M20 to M40 range.

MongoDB Atlas pricing compared to self-hosting MongoDB

MongoDB Community Server is free and self-hostable, so the comparison is a per-hour cluster rate against the cost of a server you may already run. An M30 at $387.62 a month buys 8 GB of RAM. A cloud server with 8 GB of RAM runs from roughly $24 to $48 a month on DigitalOcean, and it carries no backup or egress meters.

That comparison is real but it is not free of work, and it is worth being straight about what you give up. Atlas handles automated backups, point in time restore, failover across nodes, patching and monitoring. Self-hosting means you own all of those. For a team without anyone who wants to own a database, Atlas earning its margin is a reasonable trade.

Where self-hosting genuinely wins is when you already provision servers for the application. If your API already runs on a machine you control, adding MongoDB to it costs no additional cluster fee and removes the egress charge entirely, because the application and the database are on the same host. The marginal cost is close to zero rather than $56.94 a month.

Line	Atlas M30	Self-hosted on your own server
Compute, 8 GB RAM	$387.62 a month	Cost of the server, often $24 to $48
Database licence	Included	$0, Community Server is free
Backup storage	$0.14 per GB per month	Your own storage cost
Egress to your app	$0.01 to $0.09 per GB	$0 if the app is on the same host
Failover and patching	Managed by MongoDB	You own it
Bill predictability	Varies with traffic	Flat

This is the trade DeployManage exists to make cheap. It provisions the server on your own cloud account, installs the stack, and handles deploys, so the operational cost of self-hosting is a lot closer to the managed experience than it used to be. If you are weighing the same decision for the application layer rather than the database, VPS management and managed versus unmanaged VPS cover it directly.

Is MongoDB Enterprise Advanced pricing published?

No. MongoDB does not publish a price for Enterprise Advanced and directs buyers to contact sales. That applies to the on-premise Enterprise licence, Ops Manager and the Enterprise Kubernetes Operator, none of which carry a public rate card.

We are not going to invent a number for it. Third-party pages sometimes quote per-server Enterprise figures, but MongoDB itself publishes none, so any figure you see elsewhere is either an old quote or a guess. If you need Enterprise Advanced, the price you get will depend on your negotiation rather than on a list.

What is worth knowing is that Enterprise Advanced is a different product from Atlas. Atlas is the managed cloud service priced per hour, described in full above. Enterprise Advanced is a licence for running MongoDB on your own infrastructure with commercial support, and Community Server is the free version of that same self-hosted path.