For AI agents: The documentation index is at https://docs.digitalocean.com/llms.txt. Markdown versions of pages use the same URL with index.html.md in place of the HTML page (for example, append index.html.md to the directory path instead of opening the HTML document).

You can pay for Harness Runtime with an Agent Droplets plan, your Inference and Agents balance, or your account prepayment balance. For how your balances are used, see How to Pay for Harness Runtime. There is no per-session or per-product spend limit for Harness Runtime. Usage is billed as you consume the following components.

Agent Droplets

Agent Droplets are monthly subscription plans that discount your Managed Agents usage. They are in public preview. Each month, your plan fee funds a subscription balance. Your Managed Agents usage draws from that balance until it is used up. Eligible usage is priced at your plan’s discount, and other Managed Agents usage, such as third-party models, is priced at standard rates. After the subscription balance is used up, usage continues at standard rates from your other balances, in this order: promotional credits, the Inference and Agents balance, and the account prepayment balance. For details, see How Your Usage Is Paid For.

Agent Droplets plans change how you pay for the Managed Agents resources you use. They do not add compute, models, or tools of their own.

Agent Droplets Plans

Plan fees are billed in advance, exclude taxes, and renew automatically each month until you cancel. To subscribe, upgrade, downgrade, or cancel, see Manage Your Agent Droplets Plan.

Usage Discounted by Agent Droplets

Agent Droplets plans discount the following charges:

Discounts apply to DigitalOcean’s standard list prices, the same prices charged to accounts without a plan. Serverless Inference requests you make directly, outside of Managed Agents, are not discounted.

Agent Droplets Cost Examples

The following examples show how the Pro plan’s 15% discount applies to eligible usage:

• Usage within your plan: Your eligible usage for the month comes to $40 at standard rates. You pay $34.00 at the 15% discount, all drawn from your $50 subscription balance.
• Usage beyond your plan: Your eligible usage for the month comes to $70 at standard rates. Your $50 subscription balance covers the first $58.82. The remaining $11.18 bills at standard rates from your next available balance.
• New customer with the $5 promotional credit: You subscribe to the Pro plan as a new customer, and your eligible usage for the month comes to $70 at standard rates. Your $50 subscription balance covers the first $58.82, and your $5 credit covers the next $5. The remaining $6.18 comes from your Inference and Agents balance, or from your account prepayment balance if the Inference and Agents balance is empty.

Example Usage by Plan

The following table shows estimated monthly usage for three workload types. These are the same estimates shown in the control panel. Figures are estimates, not an included allowance. Your actual usage depends on your workload.

The workload types are:

• Light: Mostly idle or waiting on human input, with lighter compute use.
• Typical: A balanced mix of inference, compute, and storage.
• Heavy: Token-intensive, with less compute and storage relative to inference.

Agent Droplets Fair Use

Agent Droplets plans allow unlimited agents, subject to fair use. DigitalOcean may apply rate or concurrency limits to maintain service quality. You cannot resell plan discounts or share them across accounts. For current limits, see Harness Runtime Limits. For full terms, see the Agent Droplets Preview Terms.

Harness Runtime 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. Agent Droplets plans discount CPU, memory, and session storage.

** Active CPU billing is coming soon. Until then, CPU is billed at 25% of the vCPUs allocated to your sandbox. Paused sessions incur no compute charges.

The following table shows the cost of one hour at full CPU utilization and peak memory for each sandbox size. Actual CPU charges follow consumption, so a session that averages less than full utilization costs less.

Pausing a session stops compute charges while preserving the workspace and conversation history, so pause sessions you are not actively using.

Public Internet Egress

Outbound traffic from the sandbox to the public internet is billed at $0.01 per GiB of egress data. Agent Droplets plans discount public internet egress.

Snapshots and Checkpoints

Snapshot and checkpoint storage is billed at $0.05 per GiB-month for the data you store. Agent Droplets plans discount snapshot and checkpoint storage.

Retained checkpoints continue to accrue storage charges while a session is paused, including when all your balances are empty.

Custom Sandbox Templates (BYOT)

Bring-your-own container images stored as custom templates are billed at $0.05 per GiB-month of image size. Agent Droplets plans discount custom template storage.

Action Gateway Tool Calls

Some tools, including Exa web search and web fetch, require prepayment and draw from your balances in the standard order as they run. Prepayment is not required for every tool. A looping or compromised agent using tools that require prepayment can drain your balances quickly. See Action Gateway Pricing for per-call rates. Agent Droplets plans discount Action Gateway tool calls.

Model Inference

When you use DigitalOcean-hosted models, model inference is billed at Inference rates. When you use external provider keys, those providers bill you directly.

Agent Droplets plans discount inference on DigitalOcean-hosted models. They do not discount third-party models, such as Anthropic Claude and OpenAI GPT, including when you use them through DigitalOcean Serverless Inference. Third-party model usage in Managed Agents still draws from your subscription balance, at standard rates.