RunPod vs Modal
for GPU workloads.
Both bill GPUs per second. RunPod rents pods and serverless workers that run your Docker image; Modal runs Python functions defined in code. The idle defaults differ more than the prices.
Facts checked against each provider's own pages
The short answer
Choose RunPod for the lowest hourly GPU prices (H100 SXM pods $4.39/hr Secure, $2.69/hr Community[1]), no egress fees on Pods[2] and Docker-image portability. Choose Modal for Python-first serverless with scale to zero by default[3] and per-second GPU prices (H100 $0.001097/s, about $3.95/hr[4]), with $30/month of free credits on Starter[4].
RunPod vs Modal on the facts
| RunPod | Modal | |
|---|---|---|
| Category | GPU cloud | Serverless GPU |
| Pricing model | Pay as you go: Pods, Serverless workers and Clusters billed by usage; storage per GB-month[1] | Plan fee plus per-second usage for GPU, CPU cores and memory[4] |
| Entry paid plan | No plans; cheapest Secure Cloud pod listed: RTX A5000 $0.27/hr; H100 PCIe $2.89/hr[1] | Team $250/mo + compute (includes $100/mo credits)[4] |
| Free tier | — | Starter: $0 + compute with $30/mo free credits, 3 seats, 1 TiB/mo free egress[4] |
| Compute billing granularity | Per second for Pods and Serverless (Serverless rounded up to the nearest second)[5] | Per second (e.g. H100 $0.001097/s; CPU $0.0000131/core/s, minimum 0.125 cores)[4] |
| Internet egress (first paid tier) | No fees for data ingress or egress on Pods[2] | $0.04/GiB beyond the plan allowance[4] |
| Included transfer | Not metered on Pods[2] | Starter 1 TiB/mo, Team 10 TiB/mo, Enterprise 100 TiB/mo[4] |
| Regions / locations | 31 global regions (also described as 30+)[1] | 9 broad and 11 narrow region codes for region selection (counted from list); pinning costs 1.15–1.75×[8] |
| Deploy model | Docker images: Pods from templates; Serverless workers run your image with a Python handler[6] | Python decorators define functions, images and endpoints; deploy with modal deploy[9] |
| GPUs | Yes: B300, B200, H200, H100 (PCIe/SXM/NVL), A100, L40S, L40, RTX 6000 Ada, L4 and consumer cards[1] | Yes, per second: B300, B200, H200, H100, RTX PRO 6000, A100 80/40 GB, L40S, A10, L4, T4[4] |
| Managed databases | — | — |
| Preview environments | — | — |
| Persistent volumes | Yes: volume disks ($0.10/GB/mo running) and network volumes ($0.07/GB/mo under 1 TB)[7] | Yes: Modal Volumes, $0.09/GiB/mo (1 TiB/mo free)[4] |
| Scale to zero | Serverless flex workers scale to zero (default idle timeout 5 s); active workers run 24/7[5] | Yes, by default; containers stay up for scaledown_window (default 60 s) after the last input[3] |
| Config file | — | None: configuration lives in Python decorators[8] |
Which to pick, by use case
| If you need… | Pick | Why |
|---|---|---|
| Long-running training or fine-tuning on one GPU | RunPod | Pods billed per second with no commitment[5], at lower hourly list prices; see GPU pricing. |
| Bursty inference that should cost nothing when idle | Either | RunPod flex workers scale to zero after a 5 s default idle timeout[5]; Modal scales to zero after a 60 s default window[3]. |
| Python teams that want infrastructure in code | Modal | Functions, images and endpoints are defined with decorators and deployed with modal deploy[9]. |
| Moving large model weights or results out | RunPod | No ingress or egress fees on Pods[2]; Modal charges $0.04/GiB beyond the plan allowance[4]. |
| Choosing the region | RunPod | RunPod lists 31 regions[1]; on Modal, pinning a region multiplies prices by 1.15–1.75×[8]. |
What you're billed for while idle
RunPod Serverless bills by the second, rounded up[5], and a flex worker waits for up to the idle timeout (default 5 seconds) before scaling down[5]; active workers run 24/7. Modal bills during application load and input processing, and by default keeps containers alive, and billed, for 60 seconds after the last input[4].
Modal's listed GPU prices are preemptible base prices; non-preemptible runs cost 3× and region selection 1.15–1.75×[4].
Moving between them
A RunPod serverless worker is a Docker image whose handler is registered with runpod.serverless.start[10]. On Modal you can bring the same image with modal.Image.from_registry[11] and expose the handler as a web endpoint[12]. The RunPod to Modal guide has the full mapping.
Run RunPod and Modal together
You don't have to pick one. AW3 Host connects both your RunPod and Modal accounts and deploys to either from one dashboard and API. Serve the same model from both, weight traffic toward the cheaper one, and fail over automatically when one runs out of capacity.
Weighted routing splits traffic between them by percentage, canary and blue-green releases promote or roll back a version in one step, and failover driven by health checks every 60 seconds moves traffic to the healthy side. Daily cost sync and budget policies show both providers' spend side by side; secrets are encrypted per environment. Free tier: 500 credits a month; Pro: $29/month.
Frequently asked questions
- Is RunPod cheaper than Modal?
- On hourly list prices, usually yes: an H100 SXM pod is $4.39/hr on RunPod Secure Cloud and $2.69/hr on Community Cloud, while Modal's H100 is $0.001097/s (about $3.95/hr) as a preemptible base price before region multipliers. Modal can cost less for bursty work because it scales to zero by default.
- Do RunPod and Modal bill per second?
- Yes. RunPod bills Pods and Serverless per second (Serverless rounded up to the nearest second). Modal bills GPU, CPU and memory per second, including load time and a 60-second default idle window.
- Does Modal charge for egress?
- Modal includes 1 TiB/month of free egress on Starter, 10 TiB on Team and 100 TiB on Enterprise, then $0.04/GiB. RunPod's docs say there are no ingress or egress fees on Pods.
Sources
- RunPod: Pricing. https://www.runpod.io/pricing (checked 2026-11-16)
- RunPod Docs: Pod pricing. https://docs.runpod.io/pods/pricing (checked 2026-11-16)
- Modal Docs: Scaling out. https://modal.com/docs/guide/scale (checked 2026-11-16)
- Modal: Pricing. https://modal.com/pricing (checked 2026-11-16)
- RunPod Docs: Serverless pricing. https://docs.runpod.io/serverless/pricing (checked 2026-11-16)
- RunPod Docs: Serverless overview. https://docs.runpod.io/serverless/overview (checked 2026-11-16)
- RunPod Docs: Network volumes. https://docs.runpod.io/storage/network-volumes (checked 2026-11-16)
- Modal Docs: Region selection. https://modal.com/docs/guide/region-selection (checked 2026-11-16)
- Modal Docs: Managing deployments. https://modal.com/docs/guide/managing-deployments (checked 2026-11-16)
- RunPod Docs: Handler functions. https://docs.runpod.io/serverless/workers/handler-functions (checked 2026-11-16)
- Modal Docs: Existing images. https://modal.com/docs/guide/existing-images (checked 2026-11-16)
- Modal Docs: Web endpoints. https://modal.com/docs/guide/webhooks (checked 2026-11-16)
Your next deploy
goes everywhere.
Connect a provider, point a service at it, and ship. Add the second provider when you're ready.
No card needed · Free tier · Pay as you go after that