Agentic AI Workflows Explained: Patterns, Infrastructure, and GPU Requirements
Agentic workflows plan, loop, and burst differently than a single model call — here's what that means for the infrastructure underneath.
All
We have some good news! We're revamping Serverless pricing to improve our user experience for individuals, startups, and enterprises. The bad news is that if you haven't moved your cloud compute workloads to Runpod yet, that decision might keep you up at night!
With new price changes, we are introducing two different types of Serverless workers to tackle many different use cases. Each worker offers additional concurrency and can handle 1 request at a time or multiple based on your use case.
- Flex Workers - These handle spikes in your workload and allow you to support higher throughputs without impacting your users. The sum of your Flex and Active workers represents the maximum throughput your Serverless endpoint can support. You can allow your endpoint to scale down to 0 by using only Flex workers.
- Active Workers - These handle consistent workloads and run 24/7 at much lower costs. Minimum workers will be updated and labeled as Active workers.
Pricing Per Second
| GPU Size | GPU Type | Flex | Active (-40%) |
|---|---|---|---|
| 16 GB | A4000 | $0.0002 | $0.00012 |
| 24 GB | A5000 | $0.00026 | $0.00016 |
| 24 GB Pro | 4090 | $0.00044 | $0.00026 |
| 48 GB | A6000 | $0.00048 | $0.00029 |
| 80 GB | A100 | $0.0013 | $0.00078 |
New vs Old Price (only Flex)
| GPU Size | GPU Type | Old | New |
|---|---|---|---|
| 16 GB | A4000 | $0.00024 | $0.0002 |
| 24 GB | A5000 | $0.00030 | $0.00026 |
| 24 GB Pro | 4090 | $0.00050 | $0.00044 |
| 48 GB | A6000 | $0.00055 | $0.00048 |
| 80 GB | A100 | $0.00140 | $0.0013 |
This change to our Serverless worker pricing (including the transition to Active and Flex workers) will go live towards the end of this month. Please reach out to us for any inquiries about Serverless at help@runpod.io.
Update:
The 40% discount on Active Workers is now live. Enjoy!
Related posts
- 16k Context LLM Models Now Available On Runpod
- Runpod Partners With Defined.ai To Democratize and Accelerate AI Development
- Worker | Local API Server Introduced with runpod-python 0.10.0
- VS Code Server | Local-Quality Development Experience
- Savings Plans Are Here For Secure Cloud Pods – How To Purchase a Monthly Plan And Save Big
- Deploy Python ML Models on Runpod, No Docker Needed
Author profile: Pardeep Singh
