S3 Standard vs Intelligent-Tiering: Which Costs Less?

Photo by Taylor Vick on Unsplash

Intelligent-Tiering is often described as “Standard, but automatically cheaper”. That is true for many buckets and false for some, and the difference is easy to calculate once you know the four inputs: GB stored, object count, average object size and how much of the data goes untouched for 30 or 90 days. This guide is for developers and engineering leads choosing a default storage class, with a dated rate table, three worked examples and a break-even rule you can apply in your head.

To run your own numbers side by side, use the free calculator to compare S3 storage class monthly cost.

How S3 Intelligent-Tiering pricing works

Every object uploaded or transitioned to Intelligent-Tiering starts in the Frequent Access tier. According to the AWS documentation on how S3 Intelligent-Tiering works, an object not accessed for 30 consecutive days moves to the Infrequent Access tier, and after 90 days of no access it moves to the Archive Instant Access tier. All three tiers return data in milliseconds. If the object is read again, it moves back to Frequent Access.

What you pay for:

• Storage at the rate of whichever tier each object is in.
• A monitoring and automation fee per object per month, for objects of 128 KB or larger.
• Requests at the same rates as S3 Standard.

What you don’t pay for: retrieval fees, tiering charges when objects move between tiers, or a minimum storage duration. Those three are exactly where Standard-IA and Glacier Instant Retrieval catch people out, and they are why Intelligent-Tiering is the safer choice when you don’t know the access pattern.

Two optional tiers, Archive Access (after 90 or more days) and Deep Archive Access (after 180 or more days), go further but need a RestoreObject call before you can read the object, which takes hours. Only turn them on if your application can wait.

S3 Standard vs Intelligent-Tiering vs Standard-IA: rate table

Prices from the AWS public price list for S3, cross-checked against the Amazon S3 pricing page.

S3 Standard and the Frequent Access tier step down to $0.022 per GB after 50 TB and $0.021 after 500 TB. A lifecycle transition from Standard into Intelligent-Tiering costs $0.01 per 1,000 objects, once.

Worked examples: which storage class costs less?

Example 1: 10 TB of mixed data that mostly goes cold

You store 10 TB (10,240 GB) in 5 million objects averaging 2 MB. Roughly 20% is read every month, 40% hasn’t been touched for 30 to 90 days and 40% for more than 90 days.

• S3 Standard: 10,240 × $0.023 = $235.52 a month
• Intelligent-Tiering:
  • Frequent Access: 2,048 GB × $0.023 = $47.10
  • Infrequent Access: 4,096 GB × $0.0125 = $51.20
  • Archive Instant Access: 4,096 GB × $0.004 = $16.38
  • Monitoring: 5,000,000 ÷ 1,000 × $0.0025 = $12.50
  • Total: $47.10 + $51.20 + $16.38 + $12.50 = $127.18 a month

Intelligent-Tiering saves $108.34 a month, about 46%, once the data has settled into its tiers. The one-time lifecycle transition of 5 million existing objects costs 5,000 × $0.01 = $50.00, recovered in the first month.

Example 2: 1 TB of 200 KB objects read every day

A thumbnail cache holds 1 TB (1,024 GB) of 200 KB images, and nearly every object is read at least once a month.

• Objects: 1,024 × 1,048,576 KB ÷ 200 KB ≈ 5,368,709
• S3 Standard: 1,024 × $0.023 = $23.55 a month
• Intelligent-Tiering: same $23.55 of storage (nothing goes cold) + 5,368.709 × $0.0025 = $13.42 monitoring = $36.97 a month

Here Intelligent-Tiering costs 57% more. The monitoring fee buys nothing because no object ever sits idle for 30 days.

Example 3: s3 storage class cost for backup data

Nightly database backups: 5 TB (5,120 GB) in 2,000 large files, rarely restored. You keep them for a year, so in steady state most files are older than 90 days; for simplicity, assume all of them are.

• S3 Standard: 5,120 × $0.023 = $117.76
• Standard-IA: 5,120 × $0.0125 = $64.00, plus $0.01 per GB for every restore
• Intelligent-Tiering (settled): 5,120 × $0.004 = $20.48 + 2 × $0.0025 = $0.005 monitoring ≈ $20.49

With few, large objects the monitoring fee is negligible, and Archive Instant Access beats Standard-IA with no retrieval fee. If you can wait hours for a restore, Glacier Flexible Retrieval or Deep Archive is cheaper still; if you restore often, compare S3 Standard-IA versus Glacier cost with the retrieval fees included.

The break-even object size

The monitoring fee is $0.0025 ÷ 1,000 = $0.0000025 per object per month. An object earns that back only when it sits in a cheaper tier:

• In Infrequent Access it saves $0.023 − $0.0125 = $0.0105 per GB-month, so break-even is $0.0000025 ÷ $0.0105 ≈ 0.000238 GB ≈ 250 KB.
• In Archive Instant Access it saves $0.023 − $0.004 = $0.019 per GB-month, so break-even is $0.0000025 ÷ $0.019 ≈ 0.000132 GB ≈ 138 KB.

Rule of thumb: objects above about 250 KB that go cold for a month or more will pay for their monitoring. Objects under 128 KB aren’t monitored, so they cost exactly what S3 Standard costs. Objects between 128 KB and 250 KB only come out ahead if they reach the 90-day tier. Because the break-even depends on the size mix, find out what you actually store before switching: the script that reports the size and object count of each S3 bucket gives you GB and objects, and dividing one by the other gives the average object size.

Choose Standard if… choose Intelligent-Tiering if…

Choose S3 Standard if most objects are read at least monthly, objects are mostly under 128 KB, or the data is short-lived. Temporary uploads created through presigned S3 upload URLs and deleted within days are a typical example.

Choose Intelligent-Tiering if the access pattern is unknown or changes, objects are mostly larger than 250 KB, and a real share of the data goes untouched for 30 days or more. It is a sensible default for data lakes, user uploads and logs you keep for months.

Choose Standard-IA or Glacier Instant Retrieval if you know the data is cold, reads are rare, and objects stay for at least the 30- or 90-day minimum. Remember to estimate S3 infrequent access retrieval fees ($0.01 per GB for Standard-IA, $0.03 for Glacier Instant Retrieval) before committing.

Storage class doesn’t change request or egress pricing much, so don’t forget those lines: see how to calculate S3 GET and PUT request costs and the guide to S3 data transfer out cost per GB.

Check your buckets before you switch

ChatWithCloud can answer the inputs for you from the terminal: “What storage classes does media-archive use?”, “How many objects and GB are in each bucket?” or “Does app-logs have a lifecycle rule?” It writes AWS SDK code, runs it locally with your AWS profile and summarizes the JSON. The guide to ask AI which S3 buckets are largest walks through those questions, and asking AI why your AWS bill went up shows how to check whether storage is really the line to optimize. To see where S3 ranks against everything else, find your most expensive AWS service with Cost Explorer.

Use a read-only profile: generated code runs without a confirmation step, so a question phrased as a change could make one. The steps to connect ChatWithCloud to an AWS profile or SSO role cover setting that up. Treat answers as a starting point and verify object counts before a migration.

Frequently asked questions

Related guides

• AWS S3 pricing calculator for storage classesCompare monthly storage, request and transfer cost for your numbers.
• What the S3 free tier covers in 2026Credits for new accounts and what older accounts pay now.
• List AWS resources with natural languageInventory buckets and other resources from the terminal.
• ChatWithCloud CLI pricing and free runs15 free runs, then monthly, yearly or lifetime plans.
• AWS S3 Data Transfer Out CostHow S3 internet egress is priced.
• How to Calculate S3 GET and PUT Request CostsPer-1,000 S3 request pricing explained.