Team-first pricing

Pay for agent workflows, not every API call

Start with 25K runs/month free — no credit card. Pro is $79/month per org (up to 3 dashboard users), not per developer seat.

What does this cost in practice?

Same agent traffic, three billing models. See what you'd pay per month on Trakr vs Langfuse — workflow runs vs observations.

Langfuse Core and Hobby use public list prices (May 2026). Hobby is free up to 50K units/mo.

Scroll horizontally to see all columns

ScenarioRuns/mo≈ Langfuse unitsLangfuse CoreLangfuse HobbyTrakr planTrakr est.
Side projectinvoice-extractor2K20K$29/moFree/moStarterFree
Production agentsupport-triage8K96K$29/mo$29/moStarterFree
Growing teamresearch-agent120K1M$117/mo$117/moPro$95/mo
  • Side project: Starter — free with workflow monitoring and loop detection on every run.Langfuse Hobby is $0 at this volume (20K units). Core column shows $29/mo if you need 90-day retention.
  • Production agent: Starter — still within the 25K run limit; cost and error email alerts stay active all month.Langfuse Hobby covers 50K units/mo; this scenario is 96K units — Core at $29/mo applies.
  • Growing team: Pro ~$95/mo total — $79 base plus ~$16 overage for 20K runs above the 100K limit ($0.0008/run).Langfuse Core estimate uses May 2026 public list price; units are observations, not runs.

Price column is the paid self-serve plan; Free plan is each vendor’s no-cost tier (not the paid row’s included volume). Units are not 1:1 — one Trakr workflow run often maps to many Langfuse observations or LangSmith traces. Compare total cost for your agent shape, not raw unit counts. Assumes ~10 LLM calls per run where not stated per scenario.

MonthlyAnnual(2 months free)
25K runs free

Starter

Solo dev validating agents in production

Free

For individual developers validating agents in production.

  • 25K runs/mo
  • 14 days retention
  • 1 user
Get started free
Most PopularFlat $79 · 3 users

Pro

Small team shipping agents

$79/mo

For teams shipping agents to production with real traffic.

  • 100K runs/mo
  • 90 days retention
  • 3 users
Get started free

Upgrade anytime in the dashboard

Scale

Multiple pipelines at volume

$249/mo

For companies running multiple agent pipelines at volume.

  • 1M runs/mo
  • 1 year retention
  • Unlimited
Get started free

Upgrade anytime in the dashboard

How we compare

Scroll horizontally to compare products

ProductPriceBilling unitFree planLoop detectionRun comparisonFramework agnostic
TrakrPro $79/mo orgWorkflow runsStarter: 25K runs / 14d
LangfuseCore $29/mo orgObservationsHobby: 50K units / 30d
LangSmithPlus ~$117/mo (3 seats)TracesDeveloper: ~5K traces / 14dLimitedLangChain-first
HeliconePro $79/mo orgRequestsHobby: 10K req / 7d

Price column is the paid self-serve plan; Free plan is each vendor’s no-cost tier (not the paid row’s included volume). Units are not 1:1 — one Trakr workflow run often maps to many Langfuse observations or LangSmith traces. Compare total cost for your agent shape, not raw unit counts.

Scroll horizontally to compare tiers

FeatureStarterProScale
Runs per month25K100K1M
Data retention14 days90 days1 year
Team members13Unlimited
Real-time cost monitoring
Loop detection
Alert rulesBasicAll typesAll types
REST read API & access tokens
Grafana / CI export quickstart
Workflow tool breakdown
Run comparison
Slack & PagerDuty
Overage billing
Priority support

Overage billing

Pro and Scale only. $0.0008 per run above limit (Pro/Scale only). Starter pauses alert notifications at the run limit — ingestion continues with no data loss.

Scroll horizontally to see all columns

PlanBase runsActual runsOverage runsOverage charge
Pro100K130K30K$24.00
Pro100K200K100K$80.00
Scale1M1M200K$160.00

At high overage levels, upgrading tiers is almost always more cost-effective than paying per-run. Pro and Scale send an automated heads-up at 80% of your base limit.

Frequently asked questions

What counts as a run?
A run is one complete agent workflow execution — from startRun() to end(). Each LLM call inside a run is tracked separately for cost, but billing counts one run per workflow execution.
How is Trakr different from Langfuse?
Trakr is built for managed workflow-level ops: one billable run per agent execution, built-in loop detection, and alert-first triage from alert to run to compare. Langfuse is stronger on evals, prompts, and self-hosted MIT deployments when you want to run your own stack.
Why not self-host Langfuse for free?
Self-hosting means operating ClickHouse, workers, and upgrades yourself — real ops time and hosting cost. Trakr is for teams that want workflow-level alerts, loop detection, and run comparison live in minutes without running observability infrastructure.
Is 25,000 runs per month enough?
Roughly 800 runs per day — enough for many solo agents and staging workloads. If you consistently exceed 25K runs or need Slack, PagerDuty, and run comparison, upgrade to Pro.
Do I need to instrument every LLM call?
No. Call Trakr.init() once and the SDK automatically patches your Anthropic and OpenAI clients. Every API call is captured without changing your existing code.
What happens when I hit my limit?
Starter pauses alert notifications at 25,000 runs per month — runs keep ingesting with no data loss. Pro and Scale allow overage at $0.0008 per additional run.
Can I cancel anytime?
Yes. Cancel from billing settings. Monthly plans end at the current Stripe period; annual plans run through the term you paid for. Data stays available until the period ends, then retention limits apply per your tier.
When do run limits reset?
Run counts reset on the 1st of each calendar month (UTC). Overage for Pro and Scale is calculated on the same schedule for the previous calendar month. Your Stripe subscription bill date may differ — base plan fees follow your Stripe anniversary, while run limits and overage follow the calendar month.
Can I export data to Grafana or CI?
Yes. Create a read access token in Settings, then use the REST quickstart for curl or cron jobs, or copy the Grafana Infinity plugin example from the dashboard. See the Dashboard API docs at /docs/api/dashboard.
What alert types are included?
Cost threshold, error rate, loop detected, and latency spike. Starter includes basic email rules for cost and error rate. Pro and Scale add Slack, PagerDuty, and all rule types.

Ready to see what your agents cost?

Get started free