gpus4all
Sign in Start free
Status

All systems operational

Last checked August 22, 2026, 23:45 UTC. We post to this page and to the mailing list when something is wrong. Uptime target is 99.5% for instances and 99.2% for the inference API over rolling 90 days.

Inference APIOperational
GPU instancesOperational
DashboardOperational
BillingOperational
Object storageOperational
Frankfurt regionOperational
Ashburn regionOperational
Singapore regionOperational
-90 days today

Uptime by day, last 90 days. The two amber bars are the 40-minute degradation on June 23 and the 5-minute network blip on April 14 recorded below. Everything else was fully green.

Uptime history

2026-08-02

Billing invoice delay (resolved)

  • Invoices for the last week of July were delayed by ~6 hours. Amounts were correct. Root cause was a cron job that lost its lock.
2026-06-23

Inference API degraded for 40 minutes (resolved)

  • Time to first token rose to ~15s for one model. One of the boxes serving it was running a nightly rebalance at the same time. Rebalance now waits for low traffic.
2026-04-14

Frankfurt instance network blip (resolved)

  • About 10 instances lost connectivity for ~5 minutes during a datacenter switch maintenance. All recovered automatically. Instances were credited one hour of compute.