Self-host Agenta on Dedicated Cloud CPU
Give your team a shared workspace for AI agents, files and scheduled jobs. Deploy Agenta on a Dedicated CPU VM, choose where agent tasks execute and connect a compatible model service.
Fluence
$2.54/mo
Hetzner
$14.00/mo
DigitalOcean
$31.50/mo
AWS
$53.50/mo
Self-host Agenta on Dedicated Cloud CPU
Give your team a shared workspace for AI agents, files and scheduled jobs. Deploy Agenta on a Dedicated CPU VM, choose where agent tasks execute and connect a compatible model service.
Fluence
$2.54/mo
Hetzner
$14.00/mo
DigitalOcean
$31.50/mo
AWS
$53.50/mo
Run Agenta as a team service
Host the web app, scheduler and background workers on a VM you manage, with persistent storage and access you configure.
Run Agenta as a team service
Host the web app, scheduler and background workers on a VM you manage, with persistent storage and access you configure.
Where the workspace, runner and models run
Host the workspace on Fluence. Use the VM’s shared runner for trusted jobs or connect Daytona for separate sandboxes.
Where the workspace, runner and models run
Host the workspace on Fluence. Use the VM’s shared runner for trusted jobs or connect Daytona for separate sandboxes.
From Linux VM to first Agenta agent run
Use Agenta’s Docker Compose deployment to get your workspace online and verify an end-to-end run.
1
Provision the VM
Select a Dedicated CPU plan and region in the Console. Attach persistent disks and configure networking and SSH access.
2
Configure the Compose stack
Install Docker with Compose and Git. Set Agenta’s public URLs, secrets, file storage and execution mode. Pin service images to the same release and start the stack.
3
Test a real workflow
Connect a model endpoint your agent harness supports. Run a job, inspect its trace and confirm that saved files survive a restart. Setup caveat: Before adding users, verify HTTPS and workspace access restrictions. Test a restore for both the database and file store.
From Linux VM to first Agenta agent run
Use Agenta’s Docker Compose deployment to get your workspace online and verify an end-to-end run.
1
Provision the VM
Select a Dedicated CPU plan and region in the Console. Attach persistent disks and configure networking and SSH access.
2
Configure the Compose stack
Install Docker with Compose and Git. Set Agenta’s public URLs, secrets, file storage and execution mode. Pin service images to the same release and start the stack.
3
Test a real workflow
Connect a model endpoint your agent harness supports. Run a job, inspect its trace and confirm that saved files survive a restart. Setup caveat: Before adding users, verify HTTPS and workspace access restrictions. Test a restore for both the database and file store.
Estimate your Agenta VM bill
Budget for the VM, disks and public IPv4 based on how long you’ll keep each resource provisioned.
Pay zero Fluence egress fees
Check the plan rate for your selected region
See compute, disk and IP charges by resource
Note: Model APIs, Daytona sandboxes, paid integrations and any separate model server have their own costs.
Estimate your Agenta VM bill
Budget for the VM, disks and public IPv4 based on how long you’ll keep each resource provisioned.
Pay zero Fluence egress fees
Check the plan rate for your selected region
See compute, disk and IP charges by resource
Note: Model APIs, Daytona sandboxes, paid integrations and any separate model server have their own costs.
Match CPU capacity to the jobs you run
A shared VPS can suit occasional agent tests. Dedicated vCPUs give you allocated capacity for recurring jobs and concurrent tool or browser sessions.
What to compare
Fluence Dedicated CPU
Shared VPS
CPU allocation
vCPUs dedicated to your VM
CPU capacity shared with other workloads
Runtime access
Control over Linux and containers
VM-level control on most VPS plans
Data persistence
Attachable persistent disks
Disk options vary by plan
Internal traffic
Regional VPC and security groups
Private networking varies by provider
Hosting charges
Compute, disk and IPv4 priced separately
Plans and resource charges vary
Outbound traffic
Zero Fluence egress fees
Transfer allowances and overages vary
Match CPU capacity to the jobs you run
A shared VPS can suit occasional agent tests. Dedicated vCPUs give you allocated capacity for recurring jobs and concurrent tool or browser sessions.
Fluence Dedicated CPU
Shared VPS
CPU allocation
vCPUs dedicated to your VM
CPU capacity shared with other workloads
Runtime access
Control over Linux and containers
VM-level control on most VPS plans
Data persistence
Attachable persistent disks
Disk options vary by plan
Internal traffic
Regional VPC and security groups
Private networking varies by provider
Hosting charges
Compute, disk and IPv4 priced separately
Plans and resource charges vary
Outbound traffic
Zero Fluence egress fees
Transfer allowances and overages vary
Manage Agenta beyond the first deployment
Manage Agenta beyond the first deployment
Workflows to build in your Agenta workspace
Workflows to build in your Agenta workspace
Pick a region for your Agenta workspace
Choose from available Fluence providers and regions based on where your team and connected services operate.
Pick a region for your Agenta workspace
Choose from available Fluence providers and regions based on where your team and connected services operate.
FAQ
What do I need to self-host Agenta on Fluence?
What VM size should I test first?
Can I run the Compose stack on a remote server?
When would I need a GPU?
Which model endpoints and tools can I connect?
Show more
FAQ
What do I need to self-host Agenta on Fluence?
What VM size should I test first?
Can I run the Compose stack on a remote server?
When would I need a GPU?
Which model endpoints and tools can I connect?
Show more
Be the first to know
Be the first to know