Lumnar Flow

Pricing

Simple, lighter than the alternatives. Bring your own AI key. Pay only for what the platform actually runs.

Free

  • 1,000 step runs / month
  • 100 flow runs / month
  • 20 scenarios max
  • Normal modules only
  • Bring your own AI key
Start free

Starter

  • 10,000 step runs / month
  • 500 flow runs / month
  • Unlimited scenarios
  • Normal modules only
Sign up

Pro

  • 50,000 step runs / month
  • 5,000 flow runs / month
  • Unlimited scenarios
  • All modules (incl. premium)
  • Vector DB, OCR, agents
  • Top-up credits available
Sign up

Scale

  • 200,000 step runs / month
  • Unlimited flow runs
  • Unlimited scenarios
  • All modules (incl. premium)
  • Priority execution queue
  • Top-up credits available
Sign up

Custom

Custom
Negotiated pricing
  • Tailored limits & SLAs
  • Unlimited scenarios
  • All premium modules
  • Priority support
  • Contact us for a quote
Contact us

Top-up credit packs Pro and above

One-time credits that never expire. Priced above plan rates — if you need top-ups every month, a plan is cheaper. Still well below Zapier and Pipedream per-step pricing.

$10
10,000 credits · $0.001/step
$40
50,000 credits · $0.0008/step
$140
200,000 credits · $0.0007/step

FAQ

What is a "step run"?

Every module execution inside a flow counts as 1 step run. A flow with three modules that runs once consumes 3 step runs and 1 flow run.

Do unused step runs roll over to the next month?

Monthly allowance resets on the 1st of each month. Top-up credits you buy separately never expire.

What counts as a premium module?

Modules that use platform-hosted infrastructure (vector DB, PDF conversion, OCR) or advanced orchestration (agents, live-chat agents). Premium modules require Pro or higher — top-up credits don't grant access.

Do all modules consume the same number of step runs?

No. Most modules (webhooks, data, HTTP, chatgpt, gmail, etc.) cost 1 step run per execution. A few heavier modules consume more because they run real compute on our servers:

  • 3× — agent, livechat_agent, livechat_responses_agent (orchestration, holds a server worker per poll)
  • 5× — qdrant (vector database operations on platform-hosted storage)
  • 10× — pdf2md (PDF → markdown conversion, CPU-heavy)
  • 20× — tesseract_ocr (OCR, the heaviest workload)

The editor's module picker shows the multiplier next to each module so you always know its cost before adding it to a flow. A 1,000-call OCR job therefore counts as 20,000 step runs — same as 20,000 webhook calls, which is fair because they cost about the same compute.

Do I need to provide my own AI key?

Yes. You connect your own OpenAI (or compatible) key. That's intentional — it means you pay your provider directly for model usage, and we only charge for platform compute. It's also why our pricing sits below task-based competitors.

How do I pay today?

Self-serve checkout is rolling out shortly. For now, get in touch and we'll upgrade your account manually in the meantime.