> ## Documentation Index
> Fetch the complete documentation index at: https://fireworks.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Regions

> Multi-region defaults and single-region H100, H200, and B200 hardware and training availability on Fireworks, plus using and changing regions and quotas.

Fireworks runs a global fleet so you can deploy models close to users, meet data-residency needs, and scale across clouds. This page covers **multi-region** (default behavior and quota groupings), **single-region** availability and hardware, how to **use and change** regions, and **quotas**.

## Multi-region (recommended)

By default, deployments are multi-region: Fireworks can move and spread them across regions as needed. Multi-regions (**GLOBAL**, **US**, **CANADA**, **EUROPE**, **APAC**) are high-level groupings of single regions. Your deployment may run in any single region(s) within that multi-region.

> [card] Flexible, elastic scaling
>
> Utilizing multiple clouds and locations maximizes the odds that there's capacity to scale.

> [card] Higher reliability
>
> Multi-region deployments enable resilience to localized outages, maintaining application availability as workloads scale across regions.

### Supported multi-regions

Supported multi-regions: `GLOBAL`, `US`, `CANADA`, `EUROPE`, `APAC`.

## Single region availability

Single regions are concrete locations (e.g. `US_IOWA_1`, `EU_FRANKFURT_1`) where your deployment can run. We have the single regions listed below available; we recommend multi-region for most users because of its advantages (elastic scaling, higher reliability). If you have a specific need for a single region, [contact us](https://fireworks.ai/contact) to request it. The table below shows which **Fireworks-managed** single regions are available and what hardware is offered in each. In the Training column, Available means training runs in that region. `--` means it does not.

| **Region** | **Accelerator Type(s)** | **Training** |
| - | - | - |
| `AP_MALAYSIA_2` | `NVIDIA_B300_288GB` | Available |
| `AP_NEWSOUTHWALES_1` | `NVIDIA_B200_180GB` | -- |
| `AP_TAIWAN_1` | `NVIDIA_GB300` | Available |
| `AP_TOKYO_1` | `NVIDIA_H100_80GB` | -- |
| `AP_TOKYO_2` | `NVIDIA_H200_141GB` | -- |
| `EU_FRANKFURT_1` | `NVIDIA_H100_80GB` | -- |
| `EU_ICELAND_1` | `NVIDIA_H200_141GB` | -- |
| `EU_ICELAND_2` | `NVIDIA_B200_180GB`, `NVIDIA_H200_141GB` | -- |
| `NA_BRITISHCOLUMBIA_1` | `NVIDIA_B300_288GB` | -- |
| `NA_BRITISHCOLUMBIA_2` | `AMD_MI350X_288GB` | -- |
| `NA_BRITISHCOLUMBIA_3` | `NVIDIA_B300_288GB` | -- |
| `US_ARIZONA_1` | `NVIDIA_H100_80GB` | -- |
| `US_ARIZONA_3` | `AMD_MI325X_256GB` | -- |
| `US_CALIFORNIA_1` | `NVIDIA_H200_141GB` | -- |
| `US_CALIFORNIA_2` | `AMD_MI325X_256GB` | -- |
| `US_GEORGIA_2` | `NVIDIA_B200_180GB` | -- |
| `US_GEORGIA_3` | `NVIDIA_H200_141GB` | -- |
| `US_ILLINOIS_1` | `NVIDIA_H100_80GB` | -- |
| `US_ILLINOIS_2` | `NVIDIA_A100_80GB` | -- |
| `US_IOWA_1` | `NVIDIA_H100_80GB` | -- |
| `US_MINNESOTA_1` | `NVIDIA_B300_288GB` | Available |
| `US_NEWYORK_1` | `AMD_MI325X_256GB` | -- |
| `US_OHIO_1` | `NVIDIA_B200_180GB` | Available |
| `US_VIRGINIA_1` | `NVIDIA_H100_80GB`, `NVIDIA_H200_141GB`, `NVIDIA_B200_180GB`, `NVIDIA_B300_288GB` | Available |
| `US_WASHINGTON_3` | `NVIDIA_B200_180GB` | -- |
| `US_WASHINGTON_4` | `NVIDIA_B200_180GB` | -- |
| `US_WASHINGTON_5` | `NVIDIA_B200_180GB` | -- |

## Using a region

When creating a deployment, you can pass the `--region` flag to pin it to a single region:

```
firectl deployment create accounts/fireworks/models/llama-v3p1-8b-instruct \
    --deployment-shape default \
    --region GLOBAL
```

## Changing regions

Updating the single region for a deployment in-place is not supported. To move a deployment to a different single region, create a new deployment in the desired region, then delete the old deployment.

## Quotas

New accounts receive GPU quota for the **GLOBAL** multi-region only. Quota is scoped per placement (multi-region or single region). The **US**, **CANADA**, **EUROPE**, and **APAC** multi-regions and all single regions start at zero quota and must be granted by Fireworks. If you deploy with a `--region` you have no quota for, creation is rejected even though that region is generally available.

To view your current quotas, run:

```
firectl quota list
```

To use single regions that are not generally available (see the table above), or to request quota for additional placements (multi-region or single region), [contact us](https://fireworks.ai/contact). To discuss Bring Your Own Cluster (BYOC) deployments, see the [BYOC overview](https://fireworks.ai/docs/ecosystem/integrations/byoc/overview.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.