Amazon S3 pricing in 2026: free tier, storage classes, Glacier and egress explained

Amazon S3 Standard costs $0.023 per GB-month in US East (N. Virginia), Glacier Deep Archive costs $0.00099 and data transfer out to the internet costs $0.09 per GB after the first 100 GB a month. The old free tier of 5 GB for 12 months is gone: new accounts get up to $200 in credits instead. The storage rate is only one line of an S3 bill, though. Requests, retrieval fees, minimum storage durations and especially downloads often decide the total, so this page goes through each line with the prices AWS publishes for US East as of October 2026, taken from the S3 pricing page and the AWS Price List. Other regions cost a little more or less, but the structure is the same.

Amazon S3 pricing at a glance

Is there an S3 free tier in 2026?

Not in the form most people search for. The “5 GB of S3 for 12 months” offer was part of what AWS now calls the legacy Free Tier, the program that applied before 15 July 2025. Its 12-month offers ran for one year from the date the account was activated, so even an account opened on 14 July 2025 saw them end in July 2026. Nobody is inside that 12-month S3 allowance any more.

Accounts created since 15 July 2025 get credits instead:

• $100 in credits when you create the account and up to $100 more for completing activities with key services.
• On the Free plan you are never charged unless you upgrade. The plan ends six months after you open the account or when the credits run out, whichever comes first, and only some services are available. S3 is one of them.
• On the Paid plan the same credits apply to your bill and usage beyond them is charged at normal rates. The S3 pricing page says Free Tier credits must be used within 12 months of the account creation date.

What did not change is the free 100 GB a month of data transfer out to the internet, which AWS aggregates across all its services. In practice, the credits are a budget you spend across AWS, not a permanent S3 allowance. If you want storage that stays free month after month, Cloudflare R2 keeps 10 GB free every month (see the Cloudflare R2 free tier) and Backblaze B2 keeps the first 10 GB free (see Backblaze B2 pricing and free tier).

S3 storage classes and their price

What stands out:

• Standard-IA costs about 46% less than Standard to keep, Glacier Instant Retrieval about 83% less and Deep Archive about 96% less. Every one of those discounts comes back as retrieval fees, pricier requests or minimum durations.
• Intelligent-Tiering moves objects for you: to Infrequent Access after 30 consecutive days without access, to Archive Instant Access after 90 days and, if you turn them on, to Archive Access after 90 days and Deep Archive Access after 180 days. There are no retrieval charges, but there is a monitoring fee of $0.0025 per 1,000 objects a month. Objects under 128 KB are not monitored and always pay the Frequent Access rate. The price list also has a prorated charge for objects deleted or overwritten before 30 days.
• Express One Zone is built for single-digit millisecond latency in one Availability Zone. It costs almost five times Standard to store, has the cheapest requests and adds small per-GB upload and retrieval charges ($0.0032 and $0.0006 per GB in the price list).

Request pricing

Every upload, list and download is a request. Prices per 1,000 requests in US East:

DELETE and CANCEL requests are free. The colder the class, the more each write costs: a PUT to Deep Archive costs ten times one to Standard, so with many small objects the request line can cancel out the storage savings. LIST is priced like a PUT, which is why tools that list large buckets again and again add up.

Amazon S3 Glacier pricing

When people talk about Glacier pricing today they usually mean three S3 storage classes with very different trade-offs:

Restores have request fees too. An Expedited restore from Flexible Retrieval costs $0.01 per request on top of the per-GB fee. From Deep Archive, Standard restores cost $0.10 per 1,000 requests and Bulk restores $0.025 per 1,000.

Two details decide whether Glacier is really cheap for you:

1. Metadata. For every object in Glacier Flexible Retrieval or Deep Archive, AWS bills 40 KB of extra metadata: 8 KB at S3 Standard rates and 32 KB at the Glacier rate. A million tiny files carry about 40 GB of billed overhead, so bundle small files into archives before you move them.
2. Early deletion. Deleting, overwriting or moving an object before 90 days (180 for Deep Archive) bills the remaining days anyway.

A quick Glacier example: 5 TB (5,120 GB) of finished projects costs about $20.48 a month in Glacier Instant Retrieval, $18.43 in Flexible Retrieval and $5.07 in Deep Archive. Pulling 500 GB back costs $15 from Instant Retrieval, $5 with a Standard restore from Flexible Retrieval or nothing in retrieval fees with a Bulk restore, plus transfer if you download it outside AWS.

Retrieval fees for the other classes

Data transfer out: the line that grows fastest

Uploading into S3 is free, and so is sending data to CloudFront. Sending it to the internet is not:

Transfer from US East (N. Virginia) to another AWS region costs from $0.01 per GB (to US East, Ohio) to $0.02 per GB (to Europe, Ireland, for example), which matters if you replicate across regions. S3 Transfer Acceleration adds its own fee, from $0.04 per GB.

At $0.09 per GB, every terabyte downloaded by the public costs about $92 in transfer alone. For comparison, Cloudflare R2 does not charge for egress at all and Backblaze B2 gives free egress up to three times what you store. The full side-by-side is in S3 vs R2 vs Spaces vs Wasabi vs B2.

Worked examples

These use US East (N. Virginia) list prices without discounts and assume the free 100 GB of transfer is not used by other AWS services.

Example 1, an app storing user content: 2 TB stored in S3 Standard, 5 million PUTs, 50 million GETs and 500 GB of downloads in a month.

Example 2, a public video library: 10 TB stored in S3 Standard, 100,000 PUTs, 5 million GETs and 20 TB of downloads.

Transfer is roughly 88% of that bill. This is the kind of workload where a provider without egress fees, or a CDN in front of S3, changes the economics completely.

Example 3, backups in Glacier Deep Archive: 50 TB stored, 1,000 PUTs a month, no restores.

The same 50 TB in S3 Standard would cost $1,177.60 a month. The catch comes on the way back: a full restore at the Standard tier is about $1,024 in retrieval fees (51,200 GB × $0.02), or about $128 with Bulk, before any transfer out if you download it outside AWS.

Costs people often miss

• Early deletion: 30 days for Standard-IA and One Zone-IA, 90 for Glacier Instant and Flexible Retrieval, 180 for Deep Archive.
• Lifecycle transitions are billed per request: $0.01 per 1,000 to Standard-IA, One Zone-IA or Intelligent-Tiering, $0.02 to Glacier Instant Retrieval and $0.05 to Deep Archive.
• Small objects in the IA and Glacier classes, because of the 128 KB minimum and the 40 KB of metadata.
• Old versions. With versioning on, every overwritten or deleted object keeps its previous versions until a lifecycle rule expires them.
• Incomplete multipart uploads keep their uploaded parts billed until you abort them or a lifecycle rule cleans them up.

When S3 is worth it, and when it is not

S3 is the right choice when your data is used by other AWS services in the same region, when you need object lock, replication or lifecycle rules across many classes or when you keep large archives you almost never read. It gets expensive when you serve a lot of data to the public internet or need a bill you can predict to the cent. For those cases, compare Cloudflare R2, Wasabi and DigitalOcean Spaces.

Managing S3 with Nubbo

The AWS Console is built for configuration, not for daily file work. Nubbo connects to Amazon S3 with an access key and its secret and gives you a file browser, share links with passwords, expiry and download limits, galleries and file requests on top of your bucket. Your S3 bill stays the same kind of bill:

• Uploads and single-file downloads go directly between your browser and S3 through presigned URLs, so you pay S3’s normal PUT, GET and transfer rates. Files from 100 MB use multipart uploads, one request per part.
• Nubbo browses from an index of file names and sizes, so it does not list the bucket on every click, but it re-lists a folder when you open it after a minute and runs periodic background syncs. Those are LIST requests.
• Multi-file ZIP downloads are assembled in your browser, which downloads each file straight from your bucket, so those bytes count as data transfer out of S3.
• Thumbnails and watermarked gallery previews are written to your own bucket under .nubbo-thumbs/, and deleted items stay in the bucket for 30 days in Nubbo’s trash. Both add a little storage.