[Fargate] [Feature Request]: Support “One-Shot” Tier for Short-Lived Containers with 10-Second Minimum Billing

### Community Note

* Please vote on this issue by adding a 👍 [reaction](https://blog.github.com/2016-03-10-add-reactions-to-pull-requests-issues-and-comments/) to the original issue to help the community and maintainers prioritize this request
* Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
* If you are interested in working on this issue or have submitted a pull request, please leave a comment

**Tell us about your request**
I would like AWS to introduce a new tier or billing option for Fargate called something like “One-Shot Tasks,” designed specifically for short-lived container workloads. This feature would allow tasks that run for only a few seconds (e.g. 2–10 seconds) to be billed in smaller increments (e.g. 10 seconds minimum) rather than the current 1-minute minimum billing period.

This “One-Shot” mode could have certain limits (e.g. lower CPU/memory caps, restricted ephemeral storage, etc.) to enable AWS to optimize infrastructure for ultra-fast startup and cost efficiency.

**Which service(s) is this request for?**
Primarily AWS Fargate, but potentially also ECS integrations and any pricing/billing systems connected to Fargate usage.

**Tell us about the problem you're trying to solve. What are you trying to do, and why is it hard?**
Many workloads run extremely short-lived jobs that complete in just a few seconds. Examples include:

- Running a single CLI command (e.g. ffmpeg, imagemagick, curl)
- Processing a tiny ETL transformation on a single record
- Handling a webhook payload from a third-party service
- Performing small ML inference tasks
- Running short administrative scripts or jobs in a CI/CD pipeline

Today, even if these tasks only run for 5 seconds, they’re billed for a full minute on Fargate. This makes running short-lived containers cost-prohibitive compared to solutions like AWS Lambda, which offers millisecond billing. However, some of these tasks require:
- Custom binaries or packages
- Larger ephemeral storage than Lambda supports
- Container-level isolation
- Custom runtimes or languages not available in Lambda

There’s currently no middle ground between Lambda (with runtime limits and constraints) and Fargate’s 60-second minimum billing.

The impact is significant: teams either:
- Overpay dramatically for short tasks
- Move workloads to EC2 (losing serverless benefits)
- Compromise on architecture by forcing workloads into Lambda even when containers are a better fit

A shorter billing granularity would unlock cost-efficient serverless containers for many practical use cases.

**Are you currently working around this issue?**
Today, people often:
- Run short tasks in always-on Fargate services, batching work artificially (increasing complexity and latency)
- Run these tasks in EC2 instances, manually managing servers
- Force workloads into Lambda despite runtime limitations
- Accept significant overpayment for short jobs on Fargate

None of these are ideal, and they introduce architectural complexity or unnecessary costs.

**Additional context**
A “One-Shot” Fargate option could:
- Offer lower minimum billing (e.g. 10 seconds)
- Cap resource configurations to ensure AWS can pool and optimize infrastructure for fast, short jobs
- Possibly use pre-warmed infrastructure or snapshots to reduce cold start times

This feature would bridge the gap between Lambda’s millisecond billing and Fargate’s flexibility for running arbitrary containers, enabling more workloads to go serverless cost-effectively.

**Attachments**
N/A

The request concerns AWS Fargate, with possible ECS and pricing or billing integrations; no repository files, tests, or entry points are identified. Start by reviewing the current Fargate minimum-billing behavior and the proposed resource limits, then define the billing semantics and constraints that would constitute a complete One-Shot tier.

Written by the indexing model from the issue text.