MongoDB Atlas Pricing: Flex Caps at $30, Then the Big Jump

By Aakash Verma · November 8, 2026

MongoDB Atlas pricing has three steps. The Free cluster costs $0 and holds 512 MB. A Flex cluster costs $8 to $30 a month depending on operations per second, and stops at 5 GB. A dedicated cluster starts at $0.08 an hour for an M10 on AWS (MongoDB’s own estimate is about $57 a month), and the tier MongoDB recommends for production traffic, M30, is $0.54 an hour, about $394 a month. The jump from the last Flex dollar to a production cluster is roughly thirteen times the bill.

The numbers here come from the MongoDB pricing page and the Atlas billing docs for Flex costs, cluster configuration costs and data transfer, all checked on November 8, 2026. Dedicated prices are per replica set, with the default storage included, in the region MongoDB quotes: N. Virginia on AWS, Virginia (eastus2) on Azure, Iowa on Google Cloud.

Flex: billed by the busiest hour

Flex replaced the old M2 and M5 shared tiers. You can’t create those since February 2025, and on January 22, 2026 Atlas moved every remaining M2 and M5 cluster to Flex. If you had one, your flat $9 or $25 a month became a usage bill.

Flex charges by the hour, and the rate depends on how many operations per second the cluster ran in that hour:

Ops per second	Per hour	30 days at that rate
0 to 100 (base)	$0.0110	$8.00
100 to 200	$0.0205	$15.00
200 to 300	$0.0288	$21.00
300 to 400	$0.0356	$26.00
400 to 500	$0.0411	$30.00

So the cap really is $30 a month. Data transfer out is included and unlimited, which is the nicest line on the whole price list.

The limits are what end the Flex stage, not the price. Storage stops at 5 GB, and MongoDB counts uncompressed BSON plus indexes, so a collection that looks like 3 GB in your exports can hit the wall with its indexes. Above 500 operations a second there’s no higher tier, just a hard limit: Atlas makes each connection wait a moment before it continues. An app that slows down at peak with no error anywhere is often that.

Dedicated: the hourly line

Past Flex you pay by the hour for a cluster tier, busy or idle. On AWS:

Tier	RAM / vCPU	Per hour	730 hours
M10	2 GB / 2	$0.08	$58.40
M20	4 GB / 2	$0.20	$146.00
M30	8 GB / 2	$0.54	$394.20
M40	16 GB / 4	$1.04	$759.20
M50	32 GB / 8	$2.00	$1,460.00
M60	64 GB / 16	$3.95	$2,883.50
M80	128 GB / 32	$7.30	$5,329.00

The page rounds the hourly figure, which is why the M10 row comes out at $58.40 here and about $57 in MongoDB’s own tooltip. Treat both as “about $57 to $58”.

MongoDB is honest about the small tiers. M10 and M20 run on shared, burstable CPUs, “suitable for pre-production and small, non-critical workloads”, and clusters under heavy load “may experience CPU throttling”. M30 is where the vCPUs become dedicated. If your app outgrew Flex because of real traffic, budget for M30, not M10.

The cloud you pick changes the bill more than people expect. The same M30 is $0.54 an hour on AWS, $0.51 on Azure and $0.44 on Google Cloud. That’s $73 a month between AWS and Google for an identical tier, before anything else, and it matters most when your app servers are already on Google Cloud, because then you also skip the cross-cloud transfer charges below.

Staircase of monthly MongoDB Atlas prices from Free through the five Flex steps to M10, M20 and M30 on AWS, with the 5 GB wall before the jump from $30 to $394

The lines under the cluster

Storage. Each tier includes a default size (10 GB on M10, 40 GB on M30 on AWS). If you raise it, Atlas bills the full custom amount, not the difference: the docs’ own example is an M10 raised from 10 GB to 50 GB, billed for 50 GB. Storage keeps billing while a cluster is paused. Pausing saves the hourly line, not the disk.

Backups. Cloud backup snapshots cost $0.14 to $0.19 per GB-month on AWS, $0.08 to $0.12 on Google Cloud and $0.34 to $0.65 on Azure, depending on the region. Snapshots are incremental, but a tier change or a failover can leave extra full snapshots behind, and the bill can reach the full volume size. Continuous backup (point-in-time restore) is priced separately by tier of size. Exporting a snapshot to S3 or other object storage is $0.125 per GB, on the uncompressed size.

Data transfer. Atlas doesn’t charge for data coming in. On AWS, data out to the internet is $0.09 to $0.16 per GB, between AWS regions $0.02 to $0.15, and inside one region $0 to $0.01. The expensive case in MongoDB’s own words: queries “from on-premises environments into Atlas, across cloud providers, or between continents”. An API on another cloud, or on another continent, reading a big collection on every page view is exactly that case.

Things that come and go. The BI Connector, billed as a daily uplift on the cluster, reached end of life in September 2026. If an old invoice still shows it, ask why.

A worked example (hypothetical)

A SaaS app outgrows Flex at 5 GB, moves to an M30 on AWS, keeps 40 GB of snapshots and sends 200 GB a month out to the internet:

M30, 730 hours: $394.20
Backups, 40 GB at $0.14: $5.60
Data out, 200 GB at $0.09: $18.00
About $417.80 a month, against $30 the month before.

Run your own numbers in our MongoDB Atlas pricing calculator. It has Flex by ops per second and the dedicated tiers on all three clouds.

Most of the money is in the cluster tier, so most savings come from the tier too: the right cloud, a smaller tier once the slow queries have indexes, and dev or staging clusters that don’t run 730 hours. If MongoDB is one line on a bigger bill, our free cloud cost audit is $0: send the bill lines, no logins, and a person replies by email. Or we do the work through our cloud cost optimization services: $0 up front, 50% of the verified monthly saving for 6 months, then 0%, and nothing if nothing is saved.