Axiom Pricing 2026: What You Actually Pay

Quick answer (2026): Axiom, the machine-data and observability platform at axiom.co, has a permanent free Personal plan ($0/month, up to 500 GB of data loaded, 10 GB-hours of query compute and 25 GB of storage per month, 30-day retention) and a usage-based Axiom Cloud plan at $25/month that bundles 1 TB of data loaded, 100 GB-hours of query compute and 100 GB of storage, then bills overage on graduated tiers starting at $0.12 per GB loaded and $0.20 per GB-hour queried, billed in compute credits at $1.00 each (Axiom rate card, 2026). Most small teams sit inside the free tier or the flat $25. Your real bill is set by two meters: how much data you load and how hard you query it, not by how many servers you run.

A quick disambiguation, because the search results are a mess: this teardown is about Axiom the observability platform (axiom.co). It is not axiom.ai (browser automation, from $15/month), not axiom.trade (a crypto trading app), and not the Axiom CRM. If you landed here trying to price log and event storage for your app, you are in the right place.

The 2026 price ladder

Axiom rebuilt its pricing in 2026 around three usage meters plus a small platform fee. Here is the whole thing (Axiom pricing page, 2026):

Correction, September 27, 2026. An earlier version of this teardown published the query-compute overage as a flat $0.183 per GB-hour, and described both overage rates as flat with unspecified volume discounts. Axiom publishes neither figure nor that structure. The string $0.183 appears nowhere on its pricing page or in its machine-readable rate card, and both meters are graduated across five tiers. The first billable tier of query compute is $0.20 per GB-hour. Every worked bill below has been recomputed on the published tiers, and the storage and compute-credit lines the old two-meter framing left out have been added.

On Axiom Cloud the overage rates are graduated, not flat: each rate applies only to the portion of your usage that falls inside its own tier, and both meters bill in compute credits priced at $1.00 per credit pay as you go (Axiom rate card, 2026).

Note the direction: the first billable units are the dearest. Storage is the third metered line, and Axiom charges it on compressed data at a default compression ratio of 95%, across three access tiers that get cheaper as data ages: $0.03 per GB-month for the first 60 days, $0.0163 from 60 to 120 days, and $0.00522 beyond that, with the first 100 GB of compressed storage free (Axiom rate card, 2026).

The "automatic volume discounts" the pricing page advertises are those graduated tiers. There is a second, separate discount Axiom publishes but the plan cards do not mention: you can pre-purchase compute credits on a 12, 24 or 36 month term for a lower price per credit, from $0.90 at 25,000 credits a year down to $0.53 at 2,500,000+ on a 36-month term, a 47% cut. That lever only opens at roughly $25,000 a year of committed compute, so it is irrelevant to every bill worked below. Enterprise features are unbundled and priced a la carte rather than locked behind a sales call: SAML SSO is $100/month, role-based access control is $50/month, audit logs are $50/month and directory sync is $100/month (Axiom, 2026). That unbundling is the useful part, and Axiom explains the reasoning in its own pricing rework write-up.

The reframe: usage meters, not a host count

Most observability bills that hurt are host-based. You pay per monitored server, then again per GB ingested, then again per indexed event. Axiom throws that model out. In its own rate card it charges for three things plus a platform fee, and none of them is a server:

1. Data loading compute (how many GB of logs, traces and metrics you ship in).
2. Query compute (GB-hours burned when dashboards, alerts and ad-hoc searches scan that data).
3. Storage (compressed GB-months, on the three ageing tiers above).

At the volumes most teams run, the compressed footprint sits inside the free 100 GB and the first two meters are the whole invoice, which is why the bills below turn on them.

That sounds simple, and for the data-loading meter it is. The trap is the second meter. Query compute is easy to forget because it is invisible until the invoice arrives: every dashboard that auto-refreshes, every alert that evaluates on a schedule, every "last 30 days" panel someone leaves open scans data and burns GB-hours. Two teams can load nearly identical volumes and get very different bills purely because one of them queries harder.

Three worked 30-day bills (2026 rates)

Assumptions are stated so you can map them to your own workload.

1. Solo side project. 200 GB loaded, 5 GB-hours of queries, 15 GB stored. Everything sits under the Personal free allowances. Bill: $0/month. The free tier is genuinely generous for hobby and early-stage apps.

2. Small SaaS in production. 700 GB loaded, 60 GB-hours of queries, 70 GB stored. All three meters land inside the Axiom Cloud included allowances (1 TB / 100 GB-hours / 100 GB). Bill: a flat $25/month. No overage, fully predictable.

3. Scaling, ingest-heavy app. 2 TB loaded, 300 GB-hours of queries, 90 GB stored.

• Base: $25
• Data loaded: 1 TB included, 1,000 GB over at $0.12 = $120
• Query compute: 100 GB-hours included, 200 over in the $0.20 tier = $40.00
• Storage: inside the free 100 GB compressed allowance = $0
• Bill: $185/month. Ingest dominates here, which is the honest, expected shape when you are shipping a lot of logs.

The query-compute trap, quantified

Now change one thing about that scaling app: keep the data roughly flat but let the dashboards and alerts run wild.

4. Query-heavy team. 1.2 TB loaded, 700 GB-hours of queries, storage inside allowance.

• Base: $25
• Data loaded: 1 TB included, 200 GB over at $0.12 = $24
• Query compute: 100 GB-hours included, then 400 in the $0.20 tier ($80.00) and 200 in the $0.16 tier ($32.00) = $112.00
• Bill: $161/month, of which $112.00 is pure query compute.

Read that again. This team loads less data than the ingest-heavy team in bill 3, yet the query line alone ($112.00) is more than four times its ingest overage ($24). The invoice is being driven by how the team looks at its data, not how much it collects. This is the single most common Axiom surprise, and it is the one the pricing page cannot warn you about because it depends entirely on your dashboard habits.

The one lever that cuts it

You do not fix bill 4 by shipping less data. You fix it by querying smarter. Take that same query-heavy team and:

• Narrow default dashboard time ranges from 30 days to 24 hours.
• Drop alert evaluation frequency where a one-minute check is not actually needed.
• Add selective APL filters so queries scan the fields and windows they need, not the whole dataset.

That typically drops query compute from 700 GB-hours to roughly 250.

• Query compute: 100 included, 150 over in the $0.20 tier = $30.00
• New bill: $79/month, down from $161.

That is an $82/month cut, roughly 51% off, with zero change to ingest and no data thrown away. Query compute is the cheapest lever on an Axiom bill because it responds to configuration, not to how much your product is doing.

Same telemetry, three different bills

The reason "which observability tool is cheapest" has no fixed answer is that each vendor meters a different thing. Point the same 2 TB/month of logs and traces at three tools and you get three shapes (all 2026 rates):

Better Stack's numbers come straight from its pricing page (2026): ingest is the whole story and querying is almost free, which inverts Axiom's shape. Datadog's pricing is the classic host-based, multi-SKU matrix; it can be fine for a handful of servers but rarely wins on price once data volume climbs. Match the meter to the workload: ingest-heavy and query-light favors Axiom's bundled terabyte, query-heavy favors Better Stack's near-free scan, and a small predictable footprint is the one place a host-based model stays competitive.

When self-hosting wins

There is a third option the pricing pages never mention: run it yourself. A Grafana Loki stack with Grafana for dashboards and Prometheus for metrics has no per-GB meter at all. You pay for object storage and compute, full stop. That flips the math once ingest is large and predictable: at multiple terabytes a month, the object-store bill can undercut any managed per-GB price.

The catch is the same one it always is. You now run Loki, its object store, retention and upgrades, and someone has to be on the hook when it falls over at 2am. Below a few TB/month the managed $25-to-$200 Axiom bill is far cheaper than the engineering time to operate your own stack. This is the standard build-versus-buy crossover: stay managed until your data volume, not your pride, makes self-hosting pencil out. If you also want error-level detail on top of raw logs, the same discipline applies to error monitoring; our Sentry pricing breakdown walks the equivalent event-volume math.

Math check

Axiom's bill is honest and predictable if you internalize one sentence: you pay for the data you load and the compute you spend querying it. The data meter tracks your product's growth, which you cannot really cheat. The query meter tracks your team's habits, which you absolutely can. Set sane default time ranges, keep alert frequencies proportionate, and the $0-to-$25 free-and-flat band covers more real apps than you would guess. When you outgrow it, watch the query line first. It is almost always the one quietly doing the damage.