Grafana pricing

About

Grafana Labs is the company behind Grafana, the open-source visualization and dashboarding tool that has become the default “single pane of glass” for infrastructure and application telemetry. Grafana Cloud is its fully-managed observability platform, built around the open-source LGTM stack (Loki for logs, Grafana for visualization, Tempo for traces, Mimir for metrics) plus Pyroscope for continuous profiling, and packages metrics, logs, traces, profiles, Kubernetes/application/frontend/database observability, synthetic and load testing (k6), incident response (IRM), and — as of 2026 — AI products (Grafana Assistant, Investigations, Agent Observability) into a single billing surface.

Grafana serves everyone from individual developers running personal projects on the free tier to large enterprises with FedRAMP, HIPAA, or Bring-Your-Own-Cloud deployment requirements. It competes directly with Datadog, Splunk, New Relic, Dynatrace, and the Elastic (ELK) stack, and its own “Compare” pages position Grafana Cloud’s pricing as more predictable and its architecture as more open (OpenTelemetry-native, no vendor lock-in) than those incumbents. Grafana Labs also continues to sell Grafana Enterprise as a self-managed, license-based product for customers who run Grafana on their own infrastructure rather than in Grafana Cloud.

Pricing summary : How Grafana’s per-product hybrid pricing works — platform fee, usage meters, and annual-commit Enterprise

Grafana uses a hybrid, per-product pricing model with three dimensions:

1. A three-tier ladder repeated per product: Free ($0), Pro (starts at a $19/month platform fee plus usage), and Enterprise (starts at a $25,000/year committed spend), independently selectable for each of 15+ products.
2. Product-specific usage meters: each product bills its own unit once the free allowance is exceeded — active metric series ($6.50/1k series), logs/traces/profiles (per-GB process/write/retain: $0.050/$0.400/$0.100 per GB), host-hours for Kubernetes/App/Database Observability ($0.0100–$0.070/host-hour), sessions for Frontend Observability ($0.750/1k sessions), active users for Visualization/Assistant/IRM ($8.00–$55.00/user), and AI tokens for Grafana Assistant/Investigations/Agent Observability ($2/1M tokens beyond allowances).
3. Automatic volume discounts: Grafana’s stated “Pricing Philosophy” (Adaptive, Scalable, Composable) means per-unit rates step down automatically as usage grows — Enterprise Metrics pricing, for example, drops “as low as $3/1k series” versus Pro’s $6.50 starting rate.

What makes this different: unlike most observability vendors that sell one bundled per-host or per-GB SKU, Grafana Cloud lets a customer turn on only the products they need and pay the $19/month platform fee (plus usage) separately for each one — so a team using only Logs and Grafana Assistant pays two platform fees and two usage meters, not one blended bill.

Pricing by product

Core telemetry — Metrics, Logs, Traces, Profiles

All four products share the same $19/mo Pro platform fee and $25,000/year Enterprise minimum commit; only the usage meter and included allowance differ.

Observability add-ons — Kubernetes, Application, Frontend, Database

All four carry the $19/mo Pro platform fee and 8x5 email support; Enterprise requires the same $25,000/year minimum commit as every other product.

AI products — Grafana Assistant, Investigations, Agent Observability

The Investigations and Agent Observability “unbilled until October 1, 2026” caveats are footnoted directly on grafana.com/pricing — do not read these as permanently free.

Testing, visualization & incident response

Sales motions across products: PLG / self-serve for Free and Pro (sign up online, no card required for Free); sales-led for Enterprise (contact-sales form, $25,000/year minimum commit, FedRAMP/HIPAA/BYOC negotiated separately).

Self-managed Grafana Enterprise (license, not Grafana Cloud)

Grafana Labs separately sells Grafana Enterprise as an on-prem/self-managed license (grafana.com/products/grafana-enterprise/) for customers who run Grafana on their own infrastructure instead of Grafana Cloud. No price is published for this SKU — it is feature-gated (SAML, LDAP/Active LDAP sync, RBAC, SCIM provisioning, team sync, dashboard insights/reporting, custom branding, critical-response SLA, long-term support, indemnification, and access to enterprise data-source plugins) against the free open-source build, and sold via the same “Get demo” / Contact Us sales motion as Grafana Cloud Enterprise.

Hidden costs : What a multi-product team actually pays

The advertised “$19/mo” Pro headline is per product — a team that turns on more than one Grafana Cloud product pays that platform fee, and its usage overage, once per product. Here’s what a mid-size team running four products together looks like, using only the rates captured in “Pricing by product” above:

Archetype — a 20-engineer team on Metrics, Logs, Grafana Assistant, and IRM

The lesson: the $19/mo Pro figure on the pricing page is a per-product platform fee, not an account-wide subscription — a team using four products pays four $19 fees before any usage overage is counted at all. Finance teams forecasting an observability budget should treat each Grafana product as its own line item rather than one platform subscription; the usage-based pricing finance guide covers the broader discipline of budgeting for a bill built from several independent usage meters.

Want to estimate your own Grafana bill? Use the Grafana pricing calculator to model your monthly cost based on active series, GB ingested, host-hours, active users, and AI tokens across the products you actually run.

Pricing evolution : From a single flat platform fee to 15+ independently metered products

Grafana Cloud’s pricing has been rebuilt at the Pro-plan level at least three times since 2022, moving each time closer to today’s fully unbundled, per-product model — a repricing cadence that mirrors the broader SaaS shift to usage-based pricing that accelerated across the industry over the same period. The company funded that iteration through five publicly reported financing events — a $220M Series C at a $3B valuation (2021), a $270M round at a $6B valuation with $250M ARR and 5,000 customers (August 2024), and a secondary transaction disclosing $400M+ ARR and 7,000 customers (September 2025), with press reporting Grafana Labs in talks at a $9B valuation by February 2026 — the classic open-core pattern of monetizing the managed/support layer on top of a free, AGPLv3-licensed core (Grafana, Loki, Tempo, Mimir).

Cadence

Tracked range: 2022-01–2026-09. Quarters not listed above were not independently verified in this research pass.

Notable changes

• 2022-04/05 — Cloud Pro repriced from a flat $49/month fee (10 users included) to per-active-user billing starting at $8/month (1 user included, additional users $8 each); confirmed via Wayback snapshots dated 2022-04-15 and 2022-05-06.
• 2023-02/03 — Cloud Pro repriced a second time to a flat $29/month bundle (5 users, 20k metrics, 100GB logs, 100GB traces, 1000 k6 VUh included), and a fixed-price ($299/month) self-serve Cloud Advanced tier launched; confirmed via Wayback snapshots dated 2023-02-05 and 2023-03-07.
• 2023-05 — Grafana IRM became the first Grafana Cloud product priced as a separate per-active-user add-on ($20/month) on top of the base Pro fee (Wayback snapshot dated 2023-05-06).
• 2023-09 — Profiles became an explicitly metered product with its own free/Pro GB allowance (Wayback snapshot dated 2023-09-24).
• 2023-10 — Frontend Observability appeared as a distinct product line (Wayback snapshot dated 2023-10-08).
• 2024-01 — Cloud Pro’s headline reframed from a flat $29/month fee to “Pay As You Go, starts at $0” (Wayback snapshot dated 2024-01-04); per-unit rates for Visualization, IRM, and k6 already matched today’s baseline, while Metrics and the Logs/Traces/Profiles GB rate did not.
• 2024-08-21 — Grafana Labs raised $270M at a $6B valuation, reporting $250M ARR and 5,000 customers (source: grafana.com/press).
• 2025-08-14 — Grafana Assistant, the company’s first AI product, launched in public preview (source: grafana.com/press).
• 2025-09-30 — Grafana Labs disclosed $400M+ ARR and 7,000 customers alongside a secondary funding transaction with new investors (source: grafana.com/press).
• 2025-10-08 — Grafana Assistant reached general availability and Grafana Assistant Investigations was introduced at ObservabilityCON 2025 (source: grafana.com/press).
• 2026-09-01 — Baseline blueprint capture confirms the fully-unbundled model: 15+ products each independently priced on Free/Pro ($19/month + usage)/Enterprise ($25,000/year commit), including Grafana Assistant, Investigations, and Agent Observability (source: grafana.com/pricing, accessed 2026-09-01).

The AGPLv3 relicensing in detail

In April 2021, Grafana Labs relicensed its core open-source projects — Grafana, Grafana Loki, and Grafana Tempo — from Apache License 2.0 to AGPLv3. The stated motivation was to stop cloud providers from offering managed versions of Grafana’s open-source projects commercially without contributing back to the project, a practice the industry press at the time called “strip-mining.” The move drew a 710-point, 559-comment Hacker News thread — one of the highest-engagement stories in the company’s history — and was covered by Computer Weekly, The Register, and InfoQ, with reception broadly positive because AGPLv3 remained an OSI-approved open-source license (unlike the source-available SSPL that Elastic and MongoDB had adopted in similar circumstances). The episode is directly relevant to Grafana’s pricing story: it defined the boundary between what stays free forever (the open-source LGTM stack) and what Grafana Labs monetizes (Grafana Cloud’s managed hosting, support, and — since 2025 — AI features), a boundary the company still leans on in its competitive compare pages against Datadog and Splunk today.

What’s unique : Distinctive mechanics in Grafana’s per-product observability pricing

Per-product platform-fee multiplication. Most hybrid-pricing vendors charge one seat or platform fee account-wide and meter usage on top of it. Grafana instead repeats the entire Free → Pro → Enterprise ladder — including the $19/month platform fee — independently for each of 15+ products. A customer who enables Metrics, Logs, Grafana Assistant, and IRM pays four separate $19 fees before a single unit of overage is counted (see Hidden costs). This is a deliberate à-la-carte design, not an accident of packaging.

The multi-meter stack mirrors Datadog’s observability pricing more than it mirrors AI-native peers. Where an AI coding tool typically has one usage dimension (tokens or requests), Grafana stacks active series, GB ingested (three separate process/write/retain rates per GB), host-hours, sessions, test executions, virtual-user-hours, and AI tokens — a structure closer to Datadog’s per-product SKU sprawl (APM, logs, RUM, Synthetics all billed separately) than to a single-metric usage model. Buyers evaluating Grafana against Datadog, Dynatrace, New Relic, or Splunk are really comparing “how many independently priced meters will I actually turn on,” not a single headline rate. Grafana’s own IRM, Visualization, and Grafana Assistant tiers price on active users specifically, a billing unit shared with contact-center and enterprise-assistant peers elsewhere in this corpus.

Adaptive Metrics and Adaptive Logs are cost governors sold as part of the product, not third-party FinOps tooling. Grafana bakes cardinality-reduction (Adaptive Metrics, “up to 80%” savings) and log-pattern-based ingest trimming (Adaptive Logs, “up to 50%”) directly into Grafana Cloud rather than leaving cost optimization to an external observability-cost vendor — an unusually candid acknowledgment that its own usage meters can otherwise run away from customers.

The October 1, 2026 unmetered on-ramp for newer AI products. Investigations and Agent Observability are both generally available today, but Grafana’s own pricing-page footnotes state their token/generation usage “isn’t billed” until October 1, 2026 — visible in the usage dashboard but not charged. This is a deliberate adoption strategy: let usage patterns normalize (and let customers get comfortable with the meter) before flipping billing on, rather than metering from day one the way Grafana Assistant itself was. It also sidesteps, for now, the value-metric problem in AI pricing — Grafana bills its AI features on active users and raw tokens rather than on any measure of investigation quality or resolved-incident value.

95th-percentile billing smooths temporary spikes but not structural cardinality growth. Grafana’s metrics meter bills on the 95th percentile of active series across the billing period, which forgives short-lived configuration spikes — but a permanently added high-cardinality label (a pod-name dimension across hundreds of Kubernetes pods, for example) still becomes a permanent line on the bill. It’s a spike-smoothing mechanic, not a cardinality-growth safety net, and the distinction matters for teams estimating cost with the usage-metric guide.

Strengths & weaknesses

Billing UX : Named cost-control tools inside Grafana Cloud’s billing surface

• Cost Management and Billing app (grafana.com/docs/grafana-cloud/platform/cost-management-and-billing/) — a centralized in-product suite for monitoring spend, attributing costs to teams/projects/departments (chargeback/showback), setting proactive usage alerts before overages occur, and downloading monthly invoices for reconciliation.
• Interactive pricing calculator — embedded directly on the pricing page (“Estimate your Grafana Cloud cost”), lets a prospect model expected monthly spend per product before signing up; every Pro row on the pricing page links to “View calculator.”
• Adaptive Metrics — an opt-in feature that eliminates unused time-series data through aggregation and cardinality optimization, advertised as cutting metrics spend “by up to 80%.”
• Adaptive Logs — the logs-equivalent control: identifies commonly-ingested log patterns and generates recommendations for dropping unused telemetry, advertised as cutting log-ingest cost “by up to 50%.”
• Per-product platform-fee gating — each product (Metrics, Logs, Assistant, IRM, etc.) is switched from Free to Pro independently, so a customer’s monthly $19 platform fee is multiplied by however many products they’ve upgraded, not charged once account-wide.

Strategic wins : What Grafana Labs got right in its pricing playbook

1. Relicensing to AGPLv3 turned an open-source threat into a trust asset

When cloud providers began offering managed Grafana on top of the Apache-2.0-licensed open-source project without contributing back, Grafana Labs relicensed Grafana, Loki, and Tempo to AGPLv3 in April 2021 rather than moving to a more restrictive source-available license the way Elastic and MongoDB had. The announcement drew a 710-point, 559-comment Hacker News thread that was largely supportive — commentators and industry press (Computer Weekly, The Register, InfoQ) explicitly contrasted it favorably with Elastic’s SSPL move, since AGPLv3 remains OSI-approved open source rather than a proprietary license dressed up as one. That reputational capital still underwrites Grafana’s credibility as an “open” alternative to Datadog and Splunk in its own compare pages today.

2. Building cost-reduction tools into the product instead of ceding that ground to third parties

Adaptive Metrics and Adaptive Logs directly attack Grafana’s own revenue (cardinality-driven metrics spend, GB-ingested logs spend) by helping customers use less of what they’re billed for. Most usage-based vendors leave this work to third-party FinOps tools; Grafana’s decision to ship it natively — and market it on the pricing page itself (“Optimize metrics by up to 80%”) — builds the kind of buyer trust that usage-based pricing guides consistently identify as a retention lever, not just a cost center.

3. A genuinely usable free tier across every product, not just the flagship

Every one of Grafana’s 15+ products — including the newest AI features — ships with its own forever-free allowance (3 AI users, 500 k6 virtual-user-hours, 50k RUM sessions, and so on) rather than gating the free tier to visualization alone. That breadth lets a prospect pilot the entire platform, not just a teaser feature, before any billing conversation starts — a stronger product-led-growth funnel than the “one free product, everything else locked” pattern common among Grafana’s enterprise-incumbent competitors.

4. Publishing the annual-commit floor instead of hiding Enterprise pricing entirely

Grafana discloses a hard $25,000/year minimum for Enterprise on its public pricing page — a rare degree of transparency for the segment. Buyers can self-qualify against that number before ever contacting sales, which shortens the sales-led motion that otherwise dominates observability-platform enterprise deals.

Areas to improve : Where Grafana’s pricing model creates friction

1. Collapse the per-product platform fee into one account-wide charge

Charging a separate $19/month platform fee for each of 15+ products is transparent in isolation but compounds badly for the multi-product customers Grafana most wants to land — the same buyer persona targeted by its bundled LGTM stack pitch. A tiered account-wide platform fee (e.g., $19 for one product, $49 for up to four, flat above that) would reward exactly the cross-sell Grafana’s own sales motion is built around, without giving up the per-product usage metering that makes the model fair at the unit-economics level.

2. Publish an Enterprise rate card, even a rough one

Grafana discloses the $25,000/year commitment floor but nothing about how per-unit rates actually scale past that point (“as low as $3/1k series” is the only concrete Enterprise number on the entire page). A published rate table — even ranges — would let prospects self-qualify further and would reduce the sales-led friction that a $25K-and-up motion otherwise imposes on mid-market buyers who don’t yet know if they’re a $30K or $300K account.

3. Give customers a hard cardinality guardrail, not just visibility tools

Adaptive Metrics and the Cost Management dashboard help customers see and trim runaway active-series counts, but neither stops a misconfigured Kubernetes label from generating thousands of billable series before a human notices. A hard, customer-set spend cap (not just an alert) on the metrics and logs meters would directly answer the recurring “unexpected bill” complaint pattern documented in third-party billing analyses of Grafana Cloud.

4. Address the product-churn narrative directly in pricing communication

The November 2025 “I can’t recommend Grafana anymore” post (245 Hacker News points, 124 comments) cited deprecated products (Grafana OnCall, Agent/Agent Flow) and breaking framework migrations as reasons to hesitate — a reputational risk that compounds with per-product pricing, since customers who’ve paid platform fees for a product that later gets deprecated or re-architected (as Mimir 3.0’s Kafka requirement did) have a direct cost complaint, not just a technical one. A public deprecation and migration-support policy tied to billing (e.g., fee credits during forced migrations) would blunt this.

Key takeaways

1. A repeatable per-product pricing ladder scales packaging faster than redesigning tiers each time. Grafana ships a new product (Grafana Assistant, Agent Observability) by dropping it onto the existing Free/Pro/Enterprise template rather than inventing new tiers — the same mechanic other multi-product platforms should copy when launching adjacent SKUs.
2. Free-tier breadth beats free-tier depth for platform-wide product-led growth. Making every product free-tier-usable (not just the flagship) lets Grafana’s self-serve funnel sell the whole platform, not one feature — a lesson for any vendor bundling a growing product line.
3. Ship your own cost-reduction tooling before a third party does. Adaptive Metrics and Adaptive Logs turn Grafana’s own usage meters into a customer-trust asset instead of a customer-suspicion liability; vendors with usage-based billing should assume a FinOps tool will exist for their product eventually and decide whether to own that experience.
4. A relicensing move framed as “stopping strip-mining” can become a permanent brand asset. The 2021 AGPLv3 shift is still cited in Grafana’s own competitive positioning five years later — proof that a well-explained, community-first licensing decision compounds in trust value long after the initial news cycle fades.
5. Independently priced meters need independently visible guardrails, not just an aggregate cost dashboard. Community complaints about Grafana Cloud bill unpredictability trace to cardinality spikes on individual meters (metrics, in particular), not the overall bill — a reminder that usage-based products need per-meter alerting, not just account-level spend visibility.

UBP implications

1. “Per-product usage metering” and “one hybrid pricing model” are not the same design decision. Grafana shows that a company can be structurally hybrid (platform fee + usage) while still choosing to multiply that hybrid unit across many independently sold products — a packaging axis orthogonal to the pricing-model axis itself, and one the broader usage-based-pricing literature rarely separates out explicitly.
2. Open-core monetization and usage-based billing are compounding, not competing, strategies. Grafana Labs monetizes the managed hosting and support layer on top of a genuinely free, AGPLv3-licensed core — the usage meters exist on the commercial layer, not the open-source layer. That separation, funded through five publicly reported rounds from $220M (2021, $3B valuation) to a reported $9B valuation in early 2026, is a template other open-source infrastructure companies (Elastic, HashiCorp, MongoDB) have converged toward from different starting licenses.
3. Committed-use pricing at a disclosed floor, not a fully custom quote, is a viable middle ground for usage-heavy enterprise deals. Publishing the $25,000/year minimum (rather than hiding all Enterprise pricing behind “contact sales”) lets Grafana qualify enterprise leads without fully exposing its per-unit discount curve — a pattern usage-based vendors selling into both SMB self-serve and enterprise sales-led motions can adapt.

Sources

• Grafana Cloud pricing page (accessed 2026-09-01)
• Grafana Cost Management and Billing docs (accessed 2026-09-01)
• Grafana Enterprise (self-managed) (accessed 2026-09-01)
• Grafana Cloud Enterprise contact-sales (accessed 2026-09-01)
• Grafana Cloud vs. competitors comparison hub (accessed 2026-09-01)
• Grafana blog (accessed 2026-09-01)
• Grafana Labs press releases (accessed 2026-09-01)
• Grafana AGPLv3 relicensing announcement (accessed 2026-09-01)

Bottom line

Grafana Cloud’s pricing is the observability industry’s most literal expression of “à la carte”: a genuinely useful free tier and a repeatable $19-plus-usage template stamped onto 15+ independently metered products — from the original LGTM telemetry stack to the newest, temporarily-unbilled AI features. It rewards teams that adopt one or two products lightly and penalizes teams that adopt many products fully, which is the direct trade-off of building an open-core observability platform on top of a beloved, AGPLv3-licensed open-source project rather than a single bundled SKU.

Want to compare Grafana against other observability pricing? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Baseline capture — Free / $19/mo Pro / $25K annual-commit Enterprise across 15+ metered products

Sep 2026

First blueprint capture of Grafana Cloud pricing: Free (10k active series, 50GB logs/traces/profiles), Pro (from $19/month platform fee + per-product usage), and Enterprise (from a $25,000/year commit) span 15+ independently metered product lines — Metrics, Logs, Traces, Profiles, Kubernetes/Application/Frontend/Database Observability, Grafana Assistant, Investigations, Agent Observability, Synthetics (API + Browser), k6 performance testing, Visualization, and IRM.

Grafana Assistant reaches GA; Investigations introduced

Oct 2025

At ObservabilityCON 2025, Grafana Assistant reached general availability and Grafana Assistant Investigations was introduced as a token-metered, agent-powered root-cause tool (source: grafana.com/press, 2025-10-08).

Grafana Assistant launches in public preview — the company's first AI product

Aug 2025

Grafana Labs launched Grafana Assistant in public preview, its first AI-powered feature and eventual first per-active-AI-user-metered product line (source: grafana.com/press, 2025-08-14).

Pro drops its flat platform-fee headline for 'Pay As You Go, starts at $0'

Jan 2024

The January 2024 Wayback snapshot shows Cloud Pro reframed from a flat $29/month fee to 'Pro Pay As You Go, starts at $0' — pay only for usage beyond the free tier, with per-unit rates (Visualization $8/active user or $55 with Enterprise Plugins, IRM $20/active user, k6 $0.15/VUh) already matching this page's 2026 baseline rates; Metrics was still $8.00/1k series (vs. $6.50 today) and Logs/Traces/Profiles were still a single $0.50/GB rate (vs. today's decomposed process/write/retain rates). The per-product $19/month platform fee seen in 2026 had not yet appeared.

Profiles becomes an explicitly metered product

Sep 2023

By the September 2023 Wayback snapshot, Profiles (continuous profiling, powered by Grafana Pyroscope) carried its own explicit free (50GB) and Pro (100GB) allowance alongside Metrics, Logs, and Traces.

Grafana IRM becomes the first product unbundled into its own add-on fee

May 2023

The May 2023 Wayback snapshot shows Grafana IRM priced separately at $20/active user/month on top of the base Pro platform fee — the first crack in the single-bundled-fee model and the ancestor of today's fully per-product pricing.

Pro repriced a second time; Cloud Advanced launches as a fixed-price self-serve tier

Mar 2023

Between the February and March 2023 Wayback snapshots, Pro moved off per-active-user pricing to a flat $29/month bundle (5 users, 20k metrics, 100GB logs, 100GB traces, 1000 k6 VUh included), and a new self-serve Cloud Advanced tier launched at a fixed $299/month — previously that step was purely custom-quoted.

Pro repriced from a flat bundle to per-active-user billing

May 2022

Between the April and May 2022 Wayback snapshots, Grafana Cloud Pro moved from a flat $49/month fee (10 users included) to per-active-user pricing starting at $8/month for 1 user (additional users $8 each), with the base metrics/logs/traces allowance reset down to 10k/50GB/50GB to match the Free tier.

Cloud Pro launches as a single flat-fee bundle

Jan 2022

Earliest tracked Wayback snapshot (January 2022): Grafana Cloud Pro billed a single flat $49/month fee covering 10 users and 15k metrics/100GB logs/100GB traces combined; Advanced was available only via a custom quote.