LangSmith Pricing
LangSmith's overage isn't one number. It's two meters that stack, and its own page never adds them up for you.
LangSmith's real numbers, verified straight from LangChain's own pricing page: Developer is $0 a month for one seat and 5,000 base traces, Plus is $39 per seat for unlimited seats and 10,000 base traces plus a bundled deployment, and Enterprise has no published price. Past the base trace allowance, two separate meters take over, a dollar fifty per LangChain Compute Unit and a dollar per LangChain Storage Unit, billed independently rather than combined into one per-trace rate. This page verifies every figure, reconciles that against the collapsed 'per 1,000 traces' number some third-party write-ups quote instead, and covers the one question no tier answers: what your own customer never sees no matter which plan you're on.
Verified against LangChain's own pricing page
What LangSmith actually costs
LangSmith's pricing runs three tiers, and every figure below comes from LangChain's own current pricing page. Developer costs $0 a month, caps at a single seat, and includes up to 5,000 base traces a month before pay-as-you-go charges apply, with community support only. It's built for one person testing an agent, not a team.
Plus costs $39 per seat a month with no cap on how many seats an account can add, includes up to 10,000 base traces a month, and bundles one complimentary small serverless deployment for hosting a LangGraph agent. Enterprise has no dollar figure published anywhere on LangChain's site, quoted only through a sales conversation, and adds self-hosted or hybrid deployment with the data plane inside a customer's own VPC, custom SSO and ABAC, role-based access control that Developer and Plus don't get at any price, SOC 2 Type II, HIPAA, and GDPR certifications, and annual invoicing under a support SLA.
Three tiers, two overage meters
The three LangSmith tiers, in plain numbers
Straight from LangChain's own pricing page, not a third-party estimate.
Developer: $0 a month
One seat maximum, 5,000 base traces a month, then pay-as-you-go. Community support only, no bundled deployment.
Plus: $39 per seat a month
Unlimited seats, 10,000 base traces a month, one complimentary small serverless deployment included on top.
Enterprise: custom pricing
No published number. Self-hosted or hybrid deployment, custom SSO and ABAC, RBAC, SOC 2/HIPAA/GDPR, annual invoicing.
The number the pricing page never collapses into one
Past the base trace allowance, two meters bill independently
Every LangSmith tier includes a base trace count, 5,000 on Developer, 10,000 on Plus, and once a workspace clears it, LangChain's own pricing page names two separate metered units rather than one blended overage rate: a dollar fifty per LangChain Compute Unit, an LCU, and a dollar per LangChain Storage Unit, an LSU. Both stack on top of the seat price and the base allowance, and both bill independently of each other, so a workload that's heavy on compute but light on stored data pays a different real rate than one that's the reverse. Nowhere on LangChain's own page does that collapse into a single "per trace" number.
That gap is exactly where independent pricing coverage tries to fill in the blank, and not always consistently. One detailed third-party cost breakdown quotes a flat $2.50 per 1,000 base traces and $5.00 per 1,000 extended traces, a clean, easy-to-budget number that simply doesn't appear anywhere on LangChain's own pricing page, which prices by compute and storage unit instead. Treat the LCU and LSU rates as the confirmed, official numbers, and any flat per-trace figure quoted elsewhere as a third-party estimate worth checking against your own workspace's actual compute-to-storage ratio before budgeting against it.
Where the real jump sits
Plus and Enterprise aren't a small step apart
Confirmed on LangChain's own tier and enterprise documentation: none of this is available on Plus at any seat count.
RBAC is Enterprise-only, at any price
Role-based access control is listed as available exclusively on Enterprise plans. Adding more Plus seats doesn't unlock it; a team that needs to restrict who inside its own account sees which project has no self-serve path to that control, only the custom-priced conversation.
Hybrid deployment and SCIM stay behind the same wall
Running the data plane inside your own VPC, SAML or OIDC single sign-on with SCIM-based provisioning, and SOC 2, HIPAA, and GDPR certifications are all Enterprise-only, alongside custom data retention and purging and a support SLA backed by annual invoicing.
What the meter doesn't show
LangSmith's traces tell your team what a run cost. They tell your customer nothing
Every number on LangSmith's pricing page, the base trace count climbing toward 5,000 or 10,000, the LCU ticking up on a compute-heavy run, the LSU ticking up on a data-heavy one, describes what it cost your own team to trace and debug an agent. Unlimited seats on Plus and RBAC on Enterprise both decide who on your own team can see that same workspace; neither adds a second, scoped view for a customer sitting outside your company, and nothing in LangChain's own tier or enterprise documentation, Developer through Enterprise, mentions isolating a result to one specific end customer. Even Enterprise's hybrid deployment changes where the trace data lives without changing who the dashboard was built for.
If you sell an AI agent inside a product other companies pay for, that gap surfaces the moment a customer asks what the agent actually did for their account this month, a question no LCU or LSU line item answers on its own. AiAgRe's Node SDK reads the same kind of agent events LangSmith already traces, through the same LangChain, LlamaIndex, or CrewAI integrations, and tags each one with an organization identity and a customer identity at the point of ingestion. That turns into white-labeled dashboard components your customer sees inside your own product, scoped so tightly that one customer's numbers never reach another customer's view. It doesn't replace LangSmith's own tracing or its evaluation tooling, and most teams run both: LangSmith for what a run cost and how it traced, AiAgRe for what that same activity was worth to the person paying for it.
When a trace count stops being the real question
Three questions LangSmith's pricing page was never built to answer
None of these show up in a base trace count, an LCU line item, or a support SLA. All three show up once real customers start asking.
Your customer asks what their account got, not how the run traced
A customer deciding whether to renew wants their own deflection rate and cost per resolution, scoped to their own traffic, not a walkthrough of a tracing workspace they can't log into. A clean run in LangSmith reassures your engineers; it says nothing to the person deciding whether to keep paying.
Unlimited seats for your team, zero seats for theirs
Plus removes the seat cap entirely and Enterprise adds RBAC on top of it, but neither adds a scoped view for a customer outside your company, because LangSmith's workspace model was built for your own engineers, not for a second tenant layer underneath them.
Two meters track cost. Nothing tracks value
An LCU and an LSU both answer what a growing workload costs to trace, in two separate units that don't combine into one number. Neither says what that traffic was worth, the same gap a fuller look at AI agent monitoring tools as a category runs into past LangSmith alone.
FAQs
LangSmith pricing: frequently asked questions
Common questions from teams pricing LangSmith against their own trace volume before they commit to a tier, or before they add a second layer on top.
How much does LangSmith actually cost?
Three tiers, verified straight from LangChain's own pricing page. Developer costs $0 a month, caps at one seat, and includes up to 5,000 base traces a month before pay-as-you-go charges apply, with community support only. Plus costs $39 per seat a month, allows unlimited seats, includes up to 10,000 base traces a month, and bundles one complimentary small serverless deployment. Enterprise has no published number, quoted only through a sales conversation, and adds self-hosted or hybrid deployment, custom SSO, ABAC, and RBAC, a support SLA, and annual invoicing.
What happens once a workspace goes over its base trace allowance?
The plan's seat price and base trace count stop covering it, and two separate meters take over: a dollar fifty per LangChain Compute Unit, called an LCU, and a dollar per LangChain Storage Unit, called an LSU. Both are billed on top of the seat price and the base allowance already included in the tier, and they stack independently rather than combining into one number. LangChain's own pricing page states both rates plainly; the harder part is that nothing on the page converts either one into a single per-trace cost, which is exactly the number most teams actually want when they're forecasting next month's bill.
Is LangSmith's overage really $2.50 per 1,000 traces, or $1.50 per Compute Unit?
Both figures show up in different places, and they aren't describing the same thing. LangChain's own pricing page states the overage rate as a dollar fifty per LCU and a dollar per LSU, two separate metered units with no published conversion into a flat per-trace price. Independent pricing write-ups, including one detailed cost breakdown, instead quote a collapsed rate of $2.50 per 1,000 base traces and $5.00 per 1,000 extended traces, treating the two stacked meters as if they were one. That third-party number may hold up as a rough estimate for a typical trace, but it isn't the rate LangChain's own page publishes, and a team budgeting seriously should run the math against the LCU and LSU rates directly rather than the collapsed figure, since the two can diverge depending on how much of a trace's cost comes from compute versus stored data. The fuller side-by-side of how this category prices usage lives on the AI agent monitoring tools comparison.
What does Plus actually add over Developer, besides more seats?
Two things Developer doesn't have: double the base trace allowance, 10,000 a month instead of 5,000, and one complimentary small serverless deployment bundled into the plan, giving a team a hosted place to run a LangGraph agent without a separate deployment purchase. Developer caps hard at one seat; Plus removes that cap entirely at $39 per seat a month, which is usually the bigger cost driver for a growing team than the trace allowance itself, since a five-person team on Plus pays $195 a month in seats alone before any base trace or overage usage is counted.
What does Enterprise add that Plus never will?
Everything Plus can't touch regardless of how much a team is willing to pay per seat: self-hosted or hybrid deployment with the data plane inside a customer's own VPC, custom single sign-on with SAML or OIDC and SCIM-based user provisioning, attribute-based access control for tag-level data policies, SOC 2 Type II, HIPAA, and GDPR certifications, custom data retention and purging, and a support SLA backed by annual invoicing. Role-based access control itself is listed as available exclusively on Enterprise plans, which means neither Developer nor Plus gets it at any price; a team that needs to restrict who inside its own account can see which project has no self-serve tier that grants that, only the custom-priced conversation.
How does LangSmith pricing compare to Langfuse, Arize, or Braintrust?
On sticker price, LangSmith's $39-a-month Plus seat sits close to Langfuse's $29-a-month Core plan and well under Braintrust's $249-a-month Pro tier, though none of the four price the same unit: LangSmith meters seats plus a base trace count plus two stacked overage units, Langfuse meters observations, Arize meters spans and ingestion volume, and Braintrust combines data volume with evaluation scores. A dollar-for-dollar read across any two of them undercounts what each one is actually charging for, especially against Braintrust's flat fee, where the real crossover point against LangSmith's per-seat pricing is worked out on the LangSmith vs Braintrust page. The full breakdown of where Arize, Portkey, and Galileo AI fit in the same category lives on the monitoring tools comparison linked above.
Does any LangSmith plan show pricing or ROI data to my own customers, and can I run AiAgRe alongside it?
No plan does, Developer through Enterprise, and yes, running both is the normal setup rather than a workaround. Every number LangSmith renders, the base trace count climbing toward 5,000 or 10,000, an LCU or LSU line on an invoice, a prompt playground result, lives inside a workspace scoped to the account that pays LangChain's own bill. Unlimited seats on Plus and RBAC or ABAC on Enterprise both govern who inside your company can see that workspace; neither adds a second, scoped view for a customer sitting outside it, and nothing in LangChain's own tier breakdown mentions isolating a result to one specific end customer. AiAgRe's Node SDK reads the same kind of underlying agent events LangSmith already traces, through the same LangChain, LlamaIndex, or CrewAI integrations, and tags each one with an organization identity and a customer identity at the point of ingestion. That turns into white-labeled dashboard components your own customer sees inside your product, showing deflection rate and cost per resolution scoped so tightly that one customer's numbers never reach another's view. Installing it doesn't touch how LangSmith traces a run or bills an LCU; the two SDKs read the same agent, one billed and displayed to your team, the other billed and displayed to the customer paying for it.
Already tracing every run through LangSmith? Now show your customer what it did.
Request access and we'll walk through how AiAgRe's embed tokens map onto the trace data LangSmith already produces.