
Build the menu once.
Every team orders from it.
Workspaces, nodes, mounts, networking, users and groups, all in the browser. Your users pick a template and launch.
Orion Admin
×
+
←
→
↻
admin.your-domain.com/networking
☆
⋮

Orion Admin
×
+
←
→
↻
admin.your-domain.com/networking
☆
⋮

Orion Admin
×
+
admin.your-domain.com/networking
⋮

One place to run the whole fleet.
Everything an admin touches, in the browser.
No kubectl for the day-to-day.





A workspace per team or project. Activate it when work starts, hibernate it when it stops.
Every node with its roles, health, pods, and resources. This clip runs faster than real time. Adding a node takes a few minutes.
Directory groups map to roles and workspace access.
Templates that fall behind their schema get flagged, and one click brings them current.
Your NFS shares, mounted where they belong.





A workspace per team or project. Activate it when work starts, hibernate it when it stops.
Every node with its roles, health, pods, and resources. This clip runs faster than real time. Adding a node takes a few minutes.
Directory groups map to roles and workspace access.
Templates that fall behind their schema get flagged, and one click brings them current.
Your NFS shares, mounted where they belong.





A workspace per team or project. Activate it when work starts, hibernate it when it stops.
Every node with its roles, health, pods, and resources. This clip runs faster than real time. Adding a node takes a few minutes.
Directory groups map to roles and workspace access.
Templates that fall behind their schema get flagged, and one click brings them current.
Your NFS shares, mounted where they belong.
ORION ADMIN · THE POOL
One pool. Everyone draws from it.
Run more work on the GPUs you already own. Your nodes stay where they are, and Orion keeps placing work where it fits as people come and go, slicing GPUs when that fits more people.
Serve every team without building snowflakes.
Workspaces you can hibernate
Hibernate a workspace to stop its workloads and hand the compute back. Whether state is kept depends on your cluster’s configuration.
Workspaces you can hibernate
Hibernate a workspace to stop its workloads and hand the compute back. Whether state is kept depends on your cluster’s configuration.
Workspaces you can hibernate
Hibernate a workspace to stop its workloads and hand the compute back. Whether state is kept depends on your cluster’s configuration.
Quotas per team
Keep every group inside its share of the cluster.
Quotas per team
Keep every group inside its share of the cluster.
Quotas per team
Keep every group inside its share of the cluster.
Identity that reaches the file
LDAP and single sign-on synced to roles, with POSIX UID and GID on every workload.
Identity that reaches the file
LDAP and single sign-on synced to roles, with POSIX UID and GID on every workload.
Identity that reaches the file
LDAP and single sign-on synced to roles, with POSIX UID and GID on every workload.
Mounts and network rules
NFS shares, ingress and egress rules, a network range per workspace, and more.
Mounts and network rules
NFS shares, ingress and egress rules, a network range per workspace, and more.
Mounts and network rules
NFS shares, ingress and egress rules, a network range per workspace, and more.
Metrics to your dashboards
Node health on the Networking page. Metrics flow to your own dashboards through OpenTelemetry.
Metrics to your dashboards
Node health on the Networking page. Metrics flow to your own dashboards through OpenTelemetry.
Metrics to your dashboards
Node health on the Networking page. Metrics flow to your own dashboards through OpenTelemetry.
Built for your environment
Customer-hosted and built to run air-gapped, installed inside your own network.
Built for your environment
Customer-hosted and built to run air-gapped, installed inside your own network.
Built for your environment
Customer-hosted and built to run air-gapped, installed inside your own network.
“Before this, our Monday morning standup started with someone spending 45 minutes pulling numbers from six different places. Now we walk in and everything is already on screen.”
RICHARD SHUSTER · SKID SLAYER, BUILT BY KINDWELL SOLUTIONS ON ORION
Same day
provisioning, down from 5 to 8 days · Skid Slayer via Kindwell
60-90s
request to running workload
Questions admins ask
Where does Orion Admin run?
+
Where does Orion Admin run?
In your own cluster. Orion is customer-hosted, and it was built to run air-gapped if your site needs that.
Where does Orion Admin run?
+
Where does Orion Admin run?
In your own cluster. Orion is customer-hosted, and it was built to run air-gapped if your site needs that.
Where does Orion Admin run?
+
Where does Orion Admin run?
In your own cluster. Orion is customer-hosted, and it was built to run air-gapped if your site needs that.
Does it work with our identity provider?
+
Does it work with our identity provider?
Yes. LDAP and single sign-on, synced to groups and roles, with POSIX identity on every workload.
Does it work with our identity provider?
+
Does it work with our identity provider?
Yes. LDAP and single sign-on, synced to groups and roles, with POSIX identity on every workload.
Does it work with our identity provider?
+
Does it work with our identity provider?
Yes. LDAP and single sign-on, synced to groups and roles, with POSIX identity on every workload.
Can we add nodes without the command line?
+
Can we add nodes without the command line?
Yes, on k3s clusters. Add the node’s user and SSH key or password as a credential, then provision it from the Networking page and watch the setup run.
Can we add nodes without the command line?
+
Can we add nodes without the command line?
Yes, on k3s clusters. Add the node’s user and SSH key or password as a credential, then provision it from the Networking page and watch the setup run.
Can we add nodes without the command line?
+
Can we add nodes without the command line?
Yes, on k3s clusters. Add the node’s user and SSH key or password as a credential, then provision it from the Networking page and watch the setup run.
Can we monitor the cluster?
+
Can we monitor the cluster?
Yes. Node health shows on the Networking page. For cluster metrics, install the kube-prometheus-stack plugin, which is also part of the Orion Essentials bundle. Managed Core Hours have their own metrics dashboards too.
Can we monitor the cluster?
+
Can we monitor the cluster?
Yes. Node health shows on the Networking page. For cluster metrics, install the kube-prometheus-stack plugin, which is also part of the Orion Essentials bundle. Managed Core Hours have their own metrics dashboards too.
Can we monitor the cluster?
+
Can we monitor the cluster?
Yes. Node health shows on the Networking page. For cluster metrics, install the kube-prometheus-stack plugin, which is also part of the Orion Essentials bundle. Managed Core Hours have their own metrics dashboards too.
How many more workloads can we run on the GPUs we own?
+
How many more workloads can we run on the GPUs we own?
It depends on your mix. Enterprise on-prem GPUs typically sit at 10 to 15% utilization. Orion shares cards through time-slicing and hibernates or shuts down idle workloads on the schedule you set, and we measure the change on your cluster during a Proof of Value.
How many more workloads can we run on the GPUs we own?
+
How many more workloads can we run on the GPUs we own?
It depends on your mix. Enterprise on-prem GPUs typically sit at 10 to 15% utilization. Orion shares cards through time-slicing and hibernates or shuts down idle workloads on the schedule you set, and we measure the change on your cluster during a Proof of Value.
How many more workloads can we run on the GPUs we own?
+
How many more workloads can we run on the GPUs we own?
It depends on your mix. Enterprise on-prem GPUs typically sit at 10 to 15% utilization. Orion shares cards through time-slicing and hibernates or shuts down idle workloads on the schedule you set, and we measure the change on your cluster during a Proof of Value.
KEEP GOING

See your whole fleet in one view.
Book a 30-minute demo with an engineer, or install Community Edition and read the docs.

See your whole fleet in one view.
Book a 30-minute demo with an engineer, or install Community Edition and read the docs.

See your whole fleet in one view.
Book a 30-minute demo with an engineer, or install Community Edition and read the docs.