---
title: "Get a pod GPU reset"
method: GET
path: "/pods/{id}/reset-gpu/{request_id}"
tags: ["Support API Key Authentication"]
---

# Get a pod GPU reset

`GET /pods/{id}/reset-gpu/{request_id}`

## Request

- Base URL: `https://lium.io/api`
- URL: `https://lium.io/api/pods/{id}/reset-gpu/{request_id}`
- Auth: one of:
  - HTTP bearer
  - API key in header `X-API-Key`

The state of a GPU reset that `POST /pods/{id}/reset-gpu` started: `running`, then `done` (with each GPU's outcome), `failed` (with `error.code`) or `timeout` when the validator did not answer in 5 minutes. Results are kept for 1 hour.

## Path parameters

- `id` string, uuid, required
- `request_id` string, required

## Response `200`

running, done, failed or timeout

- ResetPodGpuResponse: POST answers `running`; GET polls by request_id until `done`, `failed` or `timeout`.
  - `pod_id` string, uuid, required
  - `request_id` string, required
  - `status` 'running' | 'done' | 'failed' | 'timeout', required
  - `ok` boolean, nullable
  - `container_restarted` boolean, nullable
  - `gpus` PodGpuResetOutcome[]
    - `index` integer, required
    - `uuid` string, required
    - `result` 'reset' | 'latch_cleared' | 'failed', required
    - `detail` string, nullable
  - `error` GpuResetError
    - `code` string, required
    - `message` string, required

## Other responses

- `404`: Pod not found, or gpu_reset_not_found: no such request for this pod
- `422`: Validation Error
- `503`: gpu_reset_unavailable: the state cannot be read now; poll again

## Changes

- **2026-10-23** `0531bf348672`: 1 info
  - endpoint added

[Operation history](https://skmtc.dev/lium/apis/lium-backend-api/changes/pods/:id/reset-gpu/:request_id/get.md)

---

[API](https://skmtc.dev/lium/apis/lium-backend-api.md) - [All operations](https://skmtc.dev/lium/apis/lium-backend-api/llms.txt) - [OpenAPI document](https://skmtc.dev/lium/apis/lium-backend-api/revisions/d92fd6a1972b?raw)