On October 6, 2026, the Jetstream2 engineering team will begin performing routine maintenance. This may result in brief service interruptions affecting Jetstream2 GPU and Large Memory resources.

The maintenance window will be from Tuesday, October 6 from approximately 7AM–7PM ET. This maintenance is necessary to apply necessary firmware and package updates that ensure the continued stability and performance of our infrastructure.

Existing GPU and Large Memory instances will be intermittently unavailable. Standard CPU instances will not be affected.

There will also be changes to the Jetstream2 LLM Inference Service. The Llama 4 Scout model will be retired, and the Gemma 4-31B model will be introduced during this maintenance.

Quick Summary

• Window starts: October 6, 2026 @ 7AM ET
• Window ends: October 6, 2026 @ 7PM ET (estimated)
• Affected resources: GPU and Large Memory instances; Inference Service
• Unaffected resources: CPU only, Storage
• Take action: Save work prior to maintenance window

Included In This Maintenance

• All GPU and Large Memory hypervisors will be updated and rebooted
• Jetstream2 Inference Service will see retirement of Llama 4 Scout model and introduction of Gemma 4-31B model

Expected Impact

• Existing GPU instances and new launches may be intermittently unavailable as hypervisors are rebooted
• Llama 4 Scout will no longer be available

Required Action

All unshelved GPU and Large Memory instances will be restarted as hypervisors are restarted. They will be returned to their active states. To prevent potential data loss from unsaved processes, we strongly recommend that you save all work and stop any running processes before the window begins.

We appreciate your patience as we improve the Jetstream2 environment. If you have any questions or concerns, please contact Jetstream2 Support at help@jetstream-cloud.org.