> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/deployment/choosing-a-deployment.md).

# Choosing a Deployment Option

Roboflow offers several ways to run your models and [Workflows](https://docs.roboflow.com/workflows), from fully managed cloud APIs to inference on your own hardware. The right choice depends on whether your workload is real-time or bulk, how much latency you can tolerate, where your data is allowed to live, and how much infrastructure you want to manage.

Use the table to compare options at a glance, then follow the decision flow below to narrow down.

## Comparison

| Option                                                                           | Where it runs                                                                        | GPU                                                                                | Latency profile                                                                                                                                                                    | Pricing                                                                                                                                                                                              | Region                                                                                                                                          | Plan availability                                                                                                         | Best for                                                                                                                                                             |
| -------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| [**Serverless Hosted API**](/deployment/roboflow-cloud/serverless-api.md)        | Roboflow cloud (`serverless.roboflow.com`)                                           | Yes - GPU-accelerated, auto-scaling                                                | Real-time, synchronous. The first request to a model that isn't already loaded incurs a warmup of several seconds, then subsequent requests are fast while the model stays cached. | Metered per inference in credits, based on processing time. See [roboflow.com/credits](https://roboflow.com/credits) and [Serverless pricing](/deployment/roboflow-cloud/serverless-api/pricing.md). | - (base API). The [Video Streaming API](/deployment/roboflow-cloud/serverless-api/serverless-video-streaming-api.md) supports `us`, `eu`, `ap`. | -                                                                                                                         | Getting started quickly and scaling automatically for real-time single-image or Workflow inference, when the deployment device has a persistent internet connection. |
| [**Dedicated Deployments**](/deployment/roboflow-cloud/dedicated-deployments.md) | Roboflow-managed private cloud servers (`*.roboflow.cloud`)                          | CPU (`dev-cpu` / `prod-cpu`) or GPU (`dev-gpu` / `prod-gpu`)                       | Consistent, dedicated performance. Deployments auto-pause after a period of inactivity (fixed at 1 hour for `dev-cpu` / `dev-gpu`) and resume when you send a request.             | Pay-per-hour, billed in 1-minute intervals: GPU **1 credit/hour**, CPU **0.25 credit/hour**. Request-based billing available on request.                                                             | US-based data centers only.                                                                                                                     | Core and Enterprise. See [pricing](https://roboflow.com/pricing).                                                         | Predictable or production workloads, resource isolation, and large models that need GPU acceleration (e.g. Florence-2, SAM2).                                        |
| [**Batch Processing**](/deployment/roboflow-cloud/batch-processing.md)           | Roboflow cloud (infrastructure auto-provisioned per job)                             | CPU or GPU (selected per job)                                                      | Asynchronous - not suitable for real-time. Machines are provisioned when resources are available; provisioning typically takes a few minutes with no guaranteed exact start time.  | CPU and GPU rates differ; GPU jobs are faster but more expensive. See the [pricing page](https://roboflow.com/pricing).                                                                              | -                                                                                                                                               | Growth and Enterprise.                                                                                                    | Cost-effective processing of large volumes of stored images and videos with a Workflow.                                                                              |
| [**Self-Hosted Inference**](/deployment/self-hosted/self-hosted.md)              | Your own hardware - edge device, on-prem server, or your own cloud (Docker or `pip`) | CPU, CUDA GPU, or NVIDIA Jetson                                                    | Local, low latency with no network round-trip; real-time capable on adequate hardware.                                                                                             | Free and open source - you provide and manage the infrastructure. TensorRT-optimized packages for private models require Enterprise.                                                                 | Anywhere (your infrastructure).                                                                                                                 | All plans (open source). TensorRT private-model packages and licensing to run on more than one device require Enterprise. | On-prem, edge, air-gapped, or VPC-bound workloads where you want maximum control over environment and latency.                                                       |
| [**Edge (Deployment Manager)**](/deployment/edge-devices/deployment-manager.md)  | Your edge devices, remotely managed by Roboflow                                      | Depends on device (NVIDIA Jetson, x86 + NVIDIA GPU, or Roboflow-supplied hardware) | Local edge, real-time. Devices need continuous internet access for remote management and monitoring.                                                                               | - (contact sales).                                                                                                                                                                                   | Your device locations.                                                                                                                          | Enterprise only.                                                                                                          | Setting up, deploying, and monitoring a fleet of edge devices at scale.                                                                                              |

Where a value is shown as \*\* - \*\*, the figure isn't documented on the source page - [contact support](https://roboflow.com/sales) or check the [pricing page](https://roboflow.com/pricing).

## Decision flow

1. **Real-time or bulk?** If you need results as data arrives - live video, interactive apps, per-request inference - take the real-time path. If you're processing a large backlog of stored images or videos, use [Batch Processing](/deployment/roboflow-cloud/batch-processing.md).
2. **Sync or async?** Real-time inference is synchronous: you send a request and get a response. Batch Processing is asynchronous: you submit a job and collect results when it finishes, so don't depend on an exact start time.
3. **Cloud or on-prem?** If your data can leave your network and you want zero infrastructure to manage, use a Roboflow-hosted cloud option. If your workload must stay on-prem, on the edge, or air-gapped, run [Self-Hosted Inference](/deployment/self-hosted/self-hosted.md).
4. **Managed or self-managed?**
   * In the cloud, choose [Dedicated Deployments](/deployment/roboflow-cloud/dedicated-deployments.md) when you need predictable dedicated performance or a pinned GPU type; otherwise the [Serverless Hosted API](/deployment/roboflow-cloud/serverless-api.md) is the simplest default.
   * On your own hardware, use [Deployment Manager](/deployment/edge-devices/deployment-manager.md) (Enterprise) when you want Roboflow to manage and monitor a fleet of edge devices for you; otherwise manage the [Self-Hosted Inference](/deployment/self-hosted/self-hosted.md) server yourself.

## Options

* [Serverless Hosted API](/deployment/roboflow-cloud/serverless-api.md) - hosted, auto-scaling API for models and Workflows.
* [Dedicated Deployments](/deployment/roboflow-cloud/dedicated-deployments.md) - private, Roboflow-managed cloud servers with CPU or GPU.
* [Batch Processing](/deployment/roboflow-cloud/batch-processing.md) - asynchronous, cost-effective processing of stored data.
* [Self-Hosted Inference](/deployment/self-hosted/self-hosted.md) - run Roboflow Inference on your own hardware.
* [Deployment Manager](/deployment/edge-devices/deployment-manager.md) - manage models across a fleet of edge devices.

Once you've picked an option, see the [Production Readiness Checklist](/deployment/production-checklist.md) for error handling, retries, rate limits, and replica sizing.