Price on one meter: a platform fee plus production traces, with unlimited users — Team at $99 for 1M events is the anchor
Asked:
“How should I price my AI observability product?”
Recommendation built from 8 public plans across 4 competitors — Langfuse, Arize AX, Braintrust and LangSmith — with base prices from $0 to $249/month and four different billing meters, checked on official vendor pages on 2026-09-02. All prices in USD per month.
Recommended launch price card — a hypothesis to validate, not a universal answer
One billable event = one top-level production trace/run. Child spans are free, so richer instrumentation is never penalized. Paid plans include unlimited users. Budget alerts and an optional hard cap on every metered plan.
What a month costs as event volume grows
Recommended tiers with published overage rates; the flat step then gentle slope keeps the bill predictable and reconstructable from one number a customer can count.
Findings the cards cannot say alone
How to validate before changing a number
- Test the card with 10–15 design partners; ask each for expected monthly event volume and a purchase commitment at the quoted tier.
- Track win rate and gross margin per tier; adjust included volume before touching headline prices.
- Offer a 15% annual-prepay discount on Team and Growth; hold Enterprise to annual contracts with committed event pools.
- If evaluation calls carry real pass-through model costs, sell prepaid eval credits separately at cost plus a modest margin — do not hide them in trace pricing.
- Assumes a B2B SaaS AI/LLM observability product for startup-to-mid-market engineering teams; if the product is primarily enterprise governance or an evaluation suite, shift the package upward and customize.
Competitor benchmark, plan by plan