# S3 Requests-Tier1 and Requests-Tier2

AWS cost guides · Updated 17 November 2026

**Short answer:** Requests-Tier1 is the S3 usage type for PUT, COPY, POST and LIST requests (plus tagging) against S3 Standard, and Requests-Tier2 covers GET and all other requests. Both are billed per 1,000 requests, Tier1 at a higher rate. DELETE and CANCEL requests are free. A large Tier1 line almost always means many small writes or a process listing a bucket over and over.

## What each tier covers

| Usage type | Requests |
|---|---|
| `Requests-Tier1` | PUT, COPY, POST on S3 Standard objects and tags, plus LIST on any bucket |
| `Requests-Tier2` | GET, HEAD and all other non-Tier1 requests |
| `Requests-SIA-Tier1`, `-GIR-Tier1`, `-INT-Tier1` | Writes to Standard-IA, Glacier Instant Retrieval, Intelligent-Tiering |
| `Requests-Tier3` | Lifecycle transitions into Glacier Flexible Retrieval or Deep Archive, and standard restores |
| `Requests-Tier4` | Lifecycle transitions into Glacier Instant Retrieval, Intelligent-Tiering, Standard-IA or One Zone-IA |

LIST is charged at the Tier1 rate for every storage class. Browsing a bucket in the S3 console makes real GET and LIST requests and is billed the same as API calls. Prefixes follow the usual rule: `USW2-Requests-Tier1` is us-west-2, no prefix is us-east-1.

## Common causes of a high request bill

- **Polling with LIST** to discover new objects, every few seconds, from many workers.
- **Many tiny objects**: per-event JSON files, small Parquet partitions, thumbnails. Request cost scales with object count, not size.
- **Backup or sync tools** that list or HEAD the whole bucket on every run (`aws s3 sync` over millions of keys does this).
- **Lifecycle transitions of small objects**, which add a Tier3 or Tier4 request per object.
- **A hot public object** served straight from S3 rather than through a cache.

## Find the bucket and the caller

```
# Which request usage types, and how many requests, this month
aws ce get-cost-and-usage --region us-east-1 \
  --time-period Start=2026-11-01,End=2026-11-17 \
  --granularity MONTHLY --metrics UnblendedCost UsageQuantity \
  --filter '{"Dimensions":{"Key":"SERVICE","Values":["Amazon Simple Storage Service"]}}' \
  --group-by Type=DIMENSION,Key=USAGE_TYPE Type=DIMENSION,Key=OPERATION \
  --query 'ResultsByTime[0].Groups[?contains(Keys[0], `Requests`)].[Keys[0],Keys[1],Metrics.UsageQuantity.Amount,Metrics.UnblendedCost.Amount]' \
  --output table
```

The Operation column (for example `ListBucket` or `PutObject`) shows which call dominates. To tie it to a bucket, use Cost Explorer resource-level data or a cost allocation tag on buckets. To find the caller, enable server access logging or CloudTrail data events on that bucket for a short window; both add their own cost, so turn them off afterwards.

## Reduce it

1. **Replace LIST polling with S3 Event Notifications** to SQS, SNS, Lambda or EventBridge.
2. **Batch small records** into larger objects before writing; compact small files in data lakes.
3. **Use S3 Inventory** for audits and reconciliation instead of listing the bucket; it is billed per million objects listed in the report.
4. **Put CloudFront in front of frequently read objects** so repeat GETs hit the cache.
5. **Keep small objects out of transition rules**; S3 by default does not transition objects under 128 KB, and overriding that is rarely worth it.

What not to do: add retries with no backoff when S3 returns throttling errors. Each retry is another billed request.

## Questions

### What is Requests-Tier1 in S3?

The usage type for PUT, COPY, POST and LIST requests against S3 Standard (and tagging), billed per 1,000 requests.

### Are S3 DELETE requests free?

Yes. The S3 pricing page lists DELETE and CANCEL requests as free.

### Does browsing the S3 console cost money?

Yes, a little. The console makes GET and LIST requests that are billed at the normal request rates.

## When this happens away from your desk

Request charges can jump overnight when a new job starts polling a bucket. CloudConsole, a phone app for AWS, shows spend by usage type and explains Requests-Tier1 and Tier2, so a new request pattern is easy to spot.

[About the CloudConsole app](https://cloudconsole.dev/aws-console-app) · [Download on the App Store](https://apps.apple.com/app/cloudconsole-for-aws/id6835419710)

## Sources

- [Understanding your AWS billing and usage reports for Amazon S3](https://docs.aws.amazon.com/AmazonS3/latest/userguide/aws-usage-report-understand.html)
- [Amazon S3 pricing](https://aws.amazon.com/s3/pricing/)

## Related

- [What is TimedStorage-ByteHrs?](https://cloudconsole.dev/costs/s3-timedstorage-bytehrs)
- [S3 storage classes and cost](https://cloudconsole.dev/costs/s3-storage-classes-cost)
- [Is AWS CloudTrail free?](https://cloudconsole.dev/costs/cloudtrail-cost)
- [aws s3api list-objects-v2](https://cloudconsole.dev/cli/s3api-list-objects-v2)
- [aws s3 sync](https://cloudconsole.dev/cli/s3-sync)
- [Aurora I/O-Optimized vs Aurora Standard](https://cloudconsole.dev/costs/aurora-io-optimized-vs-standard)

---

Source: https://cloudconsole.dev/costs/s3-requests-tier1-tier2

CloudConsole is an independent iPhone and Android app for Amazon Web Services by Cloudconcur Ltd. It is not affiliated with Amazon Web Services.
