Laminar Pricing Explained (2026) - What You Actually Pay

Laminar has the most unusual pricing model in the observability category, and it is not a gimmick - it genuinely changes how you should budget. Two of its three billing axes are normal. The third is where a monthly forecast falls apart. Let me decode the meter, show why the number is hard to pin down, then point at two picks you can actually forecast.

Laminar is an OpenTelemetry-native, Rust-based tracing and eval platform built for AI agents, especially browser agents. Its headline strength is that it is the only platform here you can self-host in full - the whole stack, not just a piece. Hold that thought, because self-hosting is also the escape hatch from the pricing puzzle.

The pricing model, decoded

Three self-serve tiers, each billed on two axes.

Tier	Price	Data	Signals	Retention
Free	$0	1 GB	$5	7 days
Starter	$30/mo	3 GB, then $2/GB	$15, then usage-based	30 days
Pro	$150/mo	10 GB, then $1.50/GB	$50, then usage-based	6 months
Enterprise	Custom	custom	custom	-

Data by the GB is the normal axis - traces take up storage, you pay per gigabyte, the per-GB rate drops as you move up tiers. Easy.

Signals is the axis that breaks forecasting. Signals is the feature where you describe an outcome or failure in plain language and Laminar extracts structured events across your traces. That extraction costs tokens - and here is the twist, Signals are metered by the tokens spent reading your traces, not the tokens your agent spends. So your bill scales with how much you interrogate your own data, not with how much your agent runs. Laminar says it compresses traces to roughly 10% of original size to hold the cost down, but that means your number depends on a vendor compression claim plus your own query behaviour. It is the least predictable pricing in the category, and that is not an exaggeration.

Estimate your bill (and why you partly can’t)

Work what you can, then acknowledge the fog.

Free tier. 1 GB of data and $5 of Signals, 7-day retention, one seat and one project. Fine for a solo prototype.

Starter, data axis. $30/mo includes 3 GB. Say your agents generate 5 GB of traces a month - that is 3 included plus 2 extra at $2/GB, so $4 on top, giving $34/mo on the data axis. Predictable so far.

Starter, Signals axis. This is where it goes soft. You get $15 of Signals included. But how far $15 stretches depends on how often you run natural-language queries over your traces and how well Laminar’s compression holds. A team that lives in Signals debugging its agents could blow through $15 fast. A team that sets up a few Signals and leaves them could barely touch it. There is no clean multiplier to hand you here - and that is the honest problem with the model.

The practical takeaway: you can budget the data axis to the dollar. You cannot budget the Signals axis without watching your actual usage for a month first.

The escape hatch: self-host

Here is what makes Laminar’s odd pricing tolerable. Self-hosting removes the usage billing entirely. The whole platform is open-source and documented for self-hosting - Maxim and Galileo gate self-host to Enterprise, Portkey open-sources only the gateway, Laminar open-sources everything. Run it yourself and you pay for infrastructure and ops time, not Signals.

The honest caveat: the exact feature delta between the self-hosted build and the cloud is not clearly enumerated in the docs. Managed retention tiers and Signals billing are cloud constructs, so self-hosting clearly removes the usage bill, but whether every cloud feature ships in the self-host edition is unverified. Confirm your must-have features before you commit a production stack.

One more thing to weigh. Laminar is the youngest and smallest platform in this set - founded 2024, YC S24, a $3M seed. The mitigants are that it is fully open-source, so you are not stranded if the company pivots, and it has real credibility - OpenTelemetry co-creator Ben Sigelman is an angel investor, and Browser Use documents it as their integration.

Cheaper, more predictable alternatives

If the Signals meter makes you nervous and you want a bill you can forecast, two open-source platforms are easier to budget.

Langfuse is the broader, more mature open-source default. It bills “billable units” at $8 per 100k on its $29/mo Core tier, and at 1M events a month it runs about $101/mo - a single, forecastable axis, no token-based query meter. It is MIT-licensed and self-hosts free, though the v3 stack is four services to run. Full breakdown in Langfuse pricing.

Opik is the cheapest managed cloud in the category - $19/mo for 100k spans, with $5 per additional 100k, billed on a plain spans-per-month meter. The OSS build is Apache-2.0 with the full feature set self-hosted, no gates. If you want a number you can commit to on a card, Opik is the simplest.

So which one?

• You are building browser agents or long-running agents and want OTel-native tracing you can self-host completely - Laminar, self-hosted to sidestep the Signals meter, or on cloud if you will watch it for a month first.
• You want a mature open-source default with a forecastable bill - Langfuse.
• You want the cheapest managed cloud with the simplest meter - Opik at $19/mo.

For the wider field, see Laminar alternatives and Laminar vs Langfuse. Every price here was read from laminar.sh on 26 July 2026. This category ships breaking changes monthly, so we re-verify every 30 days.