---
title: "GPU Fleet Health Summary"
method: GET
path: "/api/gpu/summary"
tags: ["GPU"]
---

# GPU Fleet Health Summary

`GET /api/gpu/summary`

## Request

- Base URL: `https://api.fiber.ac` (the document declares no server; this is the origin it was published from)
- URL: `https://api.fiber.ac/api/gpu/summary`
- Auth: none declared

Fleet-wide GPU health from dcgm-exporter telemetry: reporting count, detachment (quiet failures), thermal/XID/ECC anomalies, by-model breakdown. On deployments whose gpu_telemetry carries the dcgm-exporter 4.8.x extension (xid_total, health_status/severity/category) a `health` block adds counts by DCGM health status / severity / category and xid_total sums; otherwise `health` is null. GPUs whose exporter does not emit the opt-in DCGM_EXP_GPU_HEALTH_STATUS series are counted as `unknown` (never as pass); `health_reporting_gpus` is the number of GPUs with a real DCGM health verdict.

## Response `200`

Successful Response

- unknown

---

[API](https://skmtc.dev/fiber-ac/apis/fiber-api.md) · [All operations](https://skmtc.dev/fiber-ac/apis/fiber-api/llms.txt) · [OpenAPI document](https://skmtc.dev/fiber-ac/apis/fiber-api/revisions/58267250d793?raw)