Plans and billing

What each plan includes, what counts as a Trace and an Execution, what happens when you run out, usage beyond your plan, Lucid credit, and where to see all of it.

You are billed on three things, and nothing else:

What it isWhere it goes
TracesEverything you send: each run, each span and each product event is one. A run with 27 spans is 28.Ingest
ExecutionsWork Trodo does for you: eval steps, agent steps and alert firings.Evals, agents, alerts
Lucid creditDollars Lucid spends on model calls when it runs on Trodo's models.Lucid

Allowances belong to your organization: every project in it draws from the same pool.

Plans

FreeTrialProEnterprise
Price$0$0 for 14 days$249 a month, or $2,499 a yearContract
Traces a month10,00050,000250,000Agreed
Executions a month1,00010,00050,000Agreed
Lucid credit a month$2$6$30Agreed
Data retention7 days12 days60 daysAgreed
Largest field in one run or span512 KB1 MB1 MBAgreed
Traces a minute60600600Agreed
Past the allowanceStopsStopsStops, or billed with usage beyond your plan onBilled, never stops

A trial is a fifth of Pro for 14 days, started with a trial code: Settings → Billing → Redeem a trial code. Upgrading ends the trial and starts Pro.

An annual Pro plan is billed once a year; its allowances still refresh every month.

What counts

Traces

One per row stored: a run, a span, a product event. Only what is actually stored counts, so a retried batch never pays twice for the spans it already sent.

Executions

SourceOne Execution isFree
EvalsOne step of an eval run on one trace. An eval that settles a trace with a Python check in one step costs 1; one that walks four steps costs 4.Traces the eval's filter or sample skips, tests in the editor, a step that failed on Trodo's side
AgentsOne step that ran, whether it passed or failed.The trigger, a branch not taken, a retry of the same step
AlertsOne firing: the moment an alert crosses into breach.Checking the alert every minute

Model calls inside an eval or agent step are not Executions: an LLM judge runs on your own provider key, and a Lucid step spends Lucid credit. See Evals: billing and limits.

Lucid credit

Lucid spends credit on the model calls it makes, and every call is itemised on Settings → Usage. Two kinds:

  • Plan credit arrives with each billing period and lapses at its end.
  • Purchased credit never expires. Buy it on Settings → Billing: $10, $25, $50, $100, or any amount from $5 to $1,000.

Plan credit is spent first. Connect your own model keys in Settings → Lucid setup and Lucid runs on them without touching the balance.

When you run out

Nothing is cut off halfway, and alerts never stop.

Without usage beyond your plan
TracesNew runs, spans and events are refused with HTTP 429 until the next period or an upgrade.
ExecutionsEvals stop scoring: traces are counted under Not scored · over the plan limit. Agents are refused before they start, with the reason on the run. An agent already running finishes. Alerts still fire and notify; those firings are recorded and not charged.
Lucid creditLucid answers that there is no credit left. Add credit, or connect your own model keys.

Each project's owner gets an email at 80% and at 100% of Traces and of Executions, once each per period.

Usage beyond your plan

On Pro, Settings → Billing has a Usage beyond your plan switch in the plan card.

  • Off (the default): Traces and Executions stop at the included amounts.
  • On: they keep going, and what you use past the allowance is billed weekly to the card already on your account. No minimum.

The rates are behind the info icon beside the switch, read live from Stripe. Traces are $12 per 100,000.

Trial and Free can't turn it on: upgrade to Pro. On Enterprise it is always on, because the contract says ingestion never stops.

Usage beyond your plan never lifts the payload or per-minute limits.

Billing period

Allowances refresh monthly on your account's anniversary: the day you upgraded for Pro, the contract start for Enterprise. Upgrading mid-period starts a fresh period at zero. Unused allowance does not carry over.

Where to see it

Settings → Usage, for everyone on the team:

  • The current period and when it resets.
  • Lucid spend: what Lucid cost this period, by run.
  • Traces and Executions: used, percent, and how many are left. Past the allowance a row shows how many over, and billed when usage beyond your plan is on.
  • Executions split into Evals, Agents and Alerts, with a daily trend.

Settings → Billing, for people with billing access:

  • Lucid credit: balance, plan and purchased credit, top-ups, history.
  • Your plan, renewal date and invoices. Manage billing opens Stripe to change the card or cancel.
  • Upgrade to Pro, Start free trial, and the Usage beyond your plan switch.

If a payment fails, Stripe retries it automatically. If every retry fails, the account moves to Free.

Other limits

  • Largest field. The biggest of a run's or span's input, output and attributes, measured as JSON. Over the plan's limit the run is refused with HTTP 413 naming the field. Nothing is ever truncated.
  • Traces a minute. A sliding one-minute window. Over it, the API answers HTTP 429 with Retry-After; retry after that. A refusal here never spends your monthly allowance.
  • Retention. Data older than your plan's retention is deleted daily, with everything derived from it.
  • MCP API keys need a paid plan. On Free, creating one is refused with HTTP 402.

On this page