Simple, predictable pricing for AI agent execution
Pay for what your agents actually consume, billed per second, with no infrastructure to provision or maintain.
DigitalOcean Harness Runtime
Harness Runtime combines the functionality of a lightweight microVM, built-in tools like chromium and a coding sandbox needed by agents to do work. It offers rich lifecycle APIs that persist conversational history and working state across sessions, along with pause/resume/fork semantics so that developers can control costs and adapt workflows to the nonlinear quirks of agentic work.
Harness Runtime requires a positive prepaid account balance. There's no per-session or per-product spend limit; usage draws down your team's shared balance as you consume compute, storage, and tool calls.
View the Harness Runtime product documentation to learn more, or talk to an expert.
Compute
Each session is billed per second of active compute based on actual CPU consumption** and peak memory usage for the sandbox utilized for the session. When an agent is waiting and consuming no CPU, its CPU charge falls to zero.
| Resource | Price | Billing basis |
|---|---|---|
| CPU | $0.044 per vCPU-hour | Actual CPU consumed** |
| Memory | $0.0095 per GB-hour | Peak memory used |
| Session storage (volumes) | $0.05 per GiB-month | Peak storage consumed |
Sandbox sizes
The table below shows the cost of one hour at full CPU utilization and peak memory for each sandbox size. Actual charges follow consumption, so a session that averages less than full utilization costs less.
| Shape | vCPUs | Memory | Price at full allocation |
|---|---|---|---|
| XSmall | 1 | 1 GB | $0.0535/hr |
| Small | 2 | 2 GB | $0.107/hr |
| Medium (default) | 2 | 4 GB | $0.126/hr |
| Large | 4 | 8 GB | $0.252/hr |
| XLarge | 16 | 32 GB | $1.008/hr |
Public internet egress
| Resource | Price | Billing basis |
|---|---|---|
| Public internet egress | $0.01 per GiB | Outbound data transfer from the sandbox |
Snapshots and checkpoints
| Resource | Price | Billing basis |
|---|---|---|
| Snapshots and checkpoints | $0.05 per GiB-month | Storage retained |
Retained checkpoints continue to accrue storage charges while a session is paused, including when your prepaid balance is $0.
Custom sandbox templates (BYOT)
| Resource | Price | Billing basis |
|---|---|---|
| Custom sandbox templates (BYOT) | $0.05 per GiB-month | Stored image size |
Store your own container image as a reusable template and run it as a sandbox without rebuilding around DigitalOcean's own harness.
Additional product pricing
| Product | Cost |
|---|---|
| Action Gateway tool calls | Some tools, including Exa web search and web fetch, require prepayment and draw down your shared balance as they run. See Action Gateway Pricing for per-call rates. |
| Serverless Inference | Passes through to Inference billing when using DigitalOcean-hosted models. External provider keys are billed by those providers. |
Harness Runtime specific questions
What is DigitalOcean Harness Runtime?
Harness Runtime is DigitalOcean's service for running AI coding agents — such as Codex CLI, Claude Code, and OpenCode — and general-purpose agent frameworks like LangGraph and Hermes, in isolated DigitalOcean sandboxes. It's available in Public Preview and enabled for all users.
How am I billed for DigitalOcean Harness Runtime?
You're billed per second of active compute** based on actual CPU utilization for the vCPUs and peak memory utilized for your session. When an agent is waiting and consuming no CPU, its CPU charge falls to zero. You will also be billed for storage for session volumes, snapshots, checkpoints, and templates. Usage draws down a prepaid account balance — there’s no per-session or per-product spend limit.
What happens when I pause a session?
Pausing stops compute charges while preserving the workspace and conversation history. You still pay to retain any snapshots or checkpoints while paused.
Do I need a prepaid balance to use DigitalOcean Harness Runtime?
Yes. Harness Runtime requires a positive prepaid account balance. A looping or compromised agent using tools that require prepayment can drain that balance quickly, so monitor usage if you're running autonomous workflows.
How does DigitalOcean Harness Runtime relate to Action Gateway?
When you use Action Gateway from a Harness Runtime session, compute and workspace storage are billed under Harness Runtime, while tool calls are billed under Action Gateway. See Action Gateway Pricing for details.
How does serverless inference get billed?
Inference passes through to your DigitalOcean Inference billing when you use DigitalOcean-hosted models. If you bring your own external provider key (for example, an Anthropic or OpenAI key), that provider bills you directly.
** Active CPU billing is coming soon. Until then, you will be billed at 25% of the vCPUs allocated to your sandbox. Paused sessions incur no compute charges.
Still have questions?
Have a complex setup or additional questions around pricing? Contact our sales team to get more information on DigitalOcean pricing.
Start building today
From GPU-powered inference and Kubernetes to managed databases and storage, get everything you need to build, scale, and deploy intelligent applications.
