Observability Platform Pricing: Examples & Companies

What is it

Observability Platform Pricing is pricing for LLM and ML observability platforms — tracing, evaluation, and monitoring of model behavior in production. The product ingests the telemetry an AI application throws off (traces, spans, evaluation scores, logged payloads) and the pricing follows the pipe: a free ingestion quota, metered volume above it, and a retention window that decides how long the data stays queryable.

The category converged on a remarkably uniform shape. Langfuse bills composite “units” past 50k free with graduated overage from $8 to $6 per 100k; Arize AI meters spans and ingested gigabytes on a dual-axis Pro plan at $50/month; Galileo counts whole traces (5,000 free, 50,000 on the $100 Pro tier); Comet’s Opik prices spans with a flat $19/month team account; Braintrust runs three meters — token credits, processed data, and scores — under a $0 or $249 platform fee. HoneyHive, Athina AI, Patronus AI, and PromptLayer each vary the recipe without leaving it.

Two structural facts shape every quote in this category: open source is everywhere (Arize’s Phoenix, Comet’s Opik, Langfuse’s self-host path), so managed pricing is disciplined by the self-hosting alternative; and the category is consolidating fast — ClickHouse acquired Langfuse in January 2026 and Cisco closed on Galileo in May 2026 — which makes pricing stability itself a buying criterion.

How it works

The standard bill is platform fee + (ingestion − quota) × rate, with retention and seats as packaging levers:

Worked example — dual-axis fan-out. A RAG app sending 1M spans and 25 GB a month to Arize AI Pro pays $50 base, ~$9.50 span overage (950k × ~$10/M), and ~$45 data overage (15 GB above the included 10 at ~$3/GB) — about $105/month, with the data axis, not the span count, doing most of the damage. The same payload-heavy behavior on Braintrust hits its processed-data meter at $3–4/GB, which counts every byte moving through the platform. Instrument first, then price — the usage-event tracking guide covers how to measure your own fan-out before committing to a tier.

Companies using this

9 in-corpus companies sell observability platforms: Arize AI, Athina AI, Braintrust, Comet (Opik), Galileo, HoneyHive, Langfuse, Patronus AI, and PromptLayer. All ship a free tier; seven publish self-serve rates and all nine gate their enterprise tier behind sales.

Patterns observed

The cohort prices ingestion, gives away seats, and charges for memory. Ingestion quotas ladder from generous free tiers into metered overage (Langfuse’s graduated curve, Arize’s dual axis, PromptLayer’s normalized txn at $0.003 falling to $0.002 on Team); seats are free or unlimited almost everywhere because a per-user tax would fight the meter; and retention is the quiet upsell — Patronus AI gates its entire free tier on a rolling 2-week window rather than on features, and Braintrust and Arize both double retention exactly where the paid tier starts.

The second pattern is defensive unit design. Every vendor that survived repricing simplified toward a unit buyers can audit: Comet prices Opik spans flat per account; Athina AI deliberately meters executions while ingested logs cost zero; PromptLayer folds requests, agent runs, and eval cells into one txn, trading precision for forecastability — the trade-off the usage-metric guide frames as legibility versus cost-tracking.

Counterexamples & variants

Braintrust is the in-category variant that breaks the “one ingestion meter” mold: three independent meters (token credits at $0.06/$0.40 per Mtok, processed data at $3–4/GB, scores at $1.50–2.50/1k) under a flat platform fee — closer to a cloud bill than a SaaS tier, with the processed-data line as the recurring surprise. Patronus AI breaks it the other way: no ingestion meter at all on the self-serve tier, just a hard 2-week data window, with the actual money in a quoted enterprise plan and an optional pay-as-you-go evaluation API. And Athina AI inverts the category’s core assumption — telemetry in is free, and only compute the platform itself initiates (prompt runs, eval cells, flow steps) burns credits — proof that “observability pricing” can meter work done rather than data received.

What this means for buyers vs vendors

For buyers

Run a week of production-shaped traffic through two or three free tiers and read the actual meters before choosing — the same workload registers as wildly different volumes depending on whether the unit is a trace (Galileo), a span (Arize, Comet), or a composite unit (Langfuse). Price the retention you actually need for audits, not the default; and in a consolidating category, ask what happens to your rates and your data export path if the vendor is acquired — two of these nine were, within five months.

For vendors

The category’s settled playbook is: meter ingestion, free the seats, ladder retention, and keep an open-source pressure valve. The open design question is bill-shock control — graduated curves (Langfuse) and flat team accounts (Opik) both beat raw linear overage on trust. If your true cost driver is bytes rather than events, surface it as its own published axis the way Arize does, rather than letting a hidden data meter aggregate into an invoice the buyer can’t reconstruct.