Credits

Calculate hosted inference costs, track usage, and keep SaaS workloads running with organization credits.

Credits are the shared balance your organization uses for metered workloads on Dreadnode SaaS. A hosted model call, a running sandbox, or another metered service deducts from the same balance.

What consumes credits

Workload	How usage is measured
Dreadnode-hosted inference (`dn/*`)	Successful model calls, using their input and output token rates
Sandboxes	Runtime while a sandbox is active
Hosted web search	Successful search calls
Telemetry	Ingested spans and stored telemetry data

Models that use your own provider key (BYOK), such as openai/* or anthropic/*, do not use Dreadnode credits. The provider bills those calls to your account instead.

Calculate an inference call

Each dn/* model has separate rates for input and output tokens. Rates are quoted per 1 million tokens. Convert both token counts to their dollar cost, add them together, then convert the total at 1,000 credits per $1 of inference cost:

input cost  = input tokens  / 1,000,000 × input rate
output cost = output tokens / 1,000,000 × output rate
credits     = round up((input cost + output cost) × 1,000)

Every successful hosted inference call costs at least 1 credit.

For example, dn/claude-sonnet-4-6 is priced at $3 per 1M input tokens and $15 per 1M output tokens. A call using 2,000 input tokens and 500 output tokens costs:

input:   2,000 / 1,000,000 × $3  = $0.0060
output:    500 / 1,000,000 × $15 = $0.0075
total:                                $0.0135
credits: round up($0.0135 × 1,000) = 14 credits

Use the rate shown for your selected model rather than treating the example rates as a price list. Hosted inference is charged after a successful call, using its actual input and output token counts. Failed and zero-cost calls do not consume inference credits.

Find rates and usage

Use the product surfaces for current rates and recorded usage:

Open Account Settings → Chat Models to compare model input and output rates in dollars per 1 million tokens.
Press Ctrl+K or run /models in the TUI. dn/* rows show input/output rates in credits per 1 million tokens; BYOK rows show provider rates in dollars.
Open Settings → Billing to see your remaining organization balance and low-balance warning.
In Inference Usage, choose a date range to see requests, input tokens, output tokens, credits, and cost by model. Organization owners can also inspect each member’s usage.
Use Transaction History for credit purchases, allocations, auto-refills, refunds, and other balance changes. The Billing page also summarizes sandbox and hosted web search usage.

The Billing page displays the current sandbox runtime rate. At the default SaaS rate, 1,000 credits is about 5 hours of sandbox runtime.

Keep the organization funded

Your organization has one credit balance shared by every member. Purchases and signup allocations add to that balance; metered workloads deduct from it. Purchased credits do not expire.

Organization owners can set a monthly credit limit for each member. The limit applies to that member’s sandbox runtime, hosted inference, and hosted web search. Usage that would exceed the limit is rejected, and the member’s running sandboxes are paused. Other members can continue using the organization balance.

Auto-refill

Enable auto-refill under Settings → Billing to add credits when the balance falls below a chosen threshold. Organization owners choose the threshold, refill quantity, and monthly cap. Auto-refill requires a saved payment method from an earlier purchase.

If a refill payment fails, auto-refill turns off. Update the payment method and enable it again from the Billing page.

When credits run out

A zero balance stops new dn/* inference requests and prevents sandboxes from starting or resuming. Running durable sandboxes pause so you can resume them after adding credits; running ephemeral sandboxes stop. Telemetry ingestion and package uploads also remain blocked until the balance is positive.

Because the platform calculates hosted inference usage from the completed response, the final call can cost more credits than remained before it started. The organization balance still stops at zero.