---
title: "List the regions and GPUs online providers can currently serve"
method: GET
path: "/v1/placement-options"
tags: ["Placement Options"]
---

# List the regions and GPUs online providers can currently serve

`GET /v1/placement-options`

## Response `200`

Returns the regions and GPUs available across online providers

- object
  - `regions` string[], required — Regions declared by at least one online provider and signed by the Console auditor
  - `regionProviderCounts` object, required — Audited online providers in each region, keyed by region
  - `gpus` object[], required — GPUs that audited online providers have free capacity for and would bid on, grouped by vendor
    - `vendor` string, required — Vendor as it appears in an SDL GPU attribute
    - `models` object[], required
      - `name` string, required — Model as it appears in an SDL GPU attribute
      - `memory` string[], required — Memory sizes a provider would bid on with the interface left unpinned
      - `interface` string[], required — Interfaces a provider would bid on with the memory left unpinned
      - `providerCount` integer, required — Audited providers with free capacity that would bid on this model with memory and interface unpinned
      - `availableUnits` integer, required — Free GPUs across those providers' nodes, counting on each node no more than it would bid on
      - `maxNodeFreeUnits` integer, required — Most GPUs a single node would bid on, which bounds the GPUs one replica can ask for
      - `variants` object[], required — Every memory and interface combination at least one provider would bid on, the model alone included
        - `memory` string, nullable, required — Memory size the SDL pins, or null to leave it unpinned
        - `interface` string, nullable, required — Interface the SDL pins, or null to leave it unpinned
        - `providerCount` integer, required — Audited providers with free capacity that advertise this exact GPU key
        - `availableUnits` integer, required — Free GPUs across those providers' nodes, counting on each node no more than it would bid on
        - `maxNodeFreeUnits` integer, required — Most GPUs a single node would bid on, which bounds the GPUs one replica can ask for

## Changes

- **2026-09-30** `c5e098882470` — 4 info
  - added the required property `gpus/items/models/items/availableUnits` to the response with the `200` status
  - added the required property `gpus/items/models/items/maxNodeFreeUnits` to the response with the `200` status
  - added the required property `gpus/items/models/items/variants/items/availableUnits` to the response with the `200` status
  - added the required property `gpus/items/models/items/variants/items/maxNodeFreeUnits` to the response with the `200` status
- **2026-09-25** `1665b8d07cfe` — 1 info
  - added the required property `gpus/items/models/items/variants` to the response with the `200` status
- **2026-09-24** `081b0ce482a8` — 2 info
  - added the required property `gpus/items/models/items/providerCount` to the response with the `200` status
  - added the required property `regionProviderCounts` to the response with the `200` status
- **2026-09-23** `c6138289fb17` — 1 info
  - endpoint added

[Change history](https://skmtc.dev/akash-network/apis/akash-network-console-api/changes/v1/placement-options/get.md)

---

[API](https://skmtc.dev/akash-network/apis/akash-network-console-api.md) · [All operations](https://skmtc.dev/akash-network/apis/akash-network-console-api/llms.txt) · [OpenAPI document](https://skmtc.dev/akash-network/apis/akash-network-console-api/revisions/11865bf7bfb4?raw)