PromptLayer Pricing
PromptLayer pricing goes from $49 a month to $500 a month in one step. Its own table never says what a unit of that $500 actually is.
PromptLayer's real numbers, verified straight from its own pricing page: Free is $0, Pro is $49, Team is $500, and Enterprise has no published price at all. Every paid tier meters three separate things, requests, agent node executions, and eval cell executions, side by side, and none of the three gets a plain-language definition anywhere on PromptLayer's site or in its docs. Between Pro and Team sits a ten times price jump with nothing published in between, a gap most seven or eight person teams fall straight into. This page verifies every figure, works out what those three meters most likely track based on PromptLayer's own product pages, and covers the one question its pricing table was never built to answer: what does any of this show the customer paying for the agent underneath it.
Verified against PromptLayer's own pricing page
What PromptLayer actually costs
PromptLayer's pricing runs four tiers, and every figure below comes from PromptLayer's own current pricing page. Free costs $0 a month, covers five users sharing one workspace, and includes twenty five hundred requests, seven hundred fifty agent node executions, and two hundred fifty eval cell executions before anything is capped outright, since Free lists no overage rate at all. Pro costs $49 a month, keeps the same five user cap and the same starting allowances as Free, but opens unlimited workspaces, unlimited saved prompts, and unlimited playground runs, and switches from a hard cap to a billed overage of $0.003 per transaction once a workspace runs past its included numbers.
Team costs $500 a month, and the jump from Pro isn't a modest step: twenty five users instead of five, a hundred thousand or more requests instead of twenty five hundred, ten thousand or more agent node executions, and the first tier to include webhooks by default, all at an overage rate that drops slightly to $0.002 per transaction. Enterprise has no dollar figure published anywhere on PromptLayer's site, quoted only through a sales conversation, and its own comparison row lists unlimited users, custom volume across every meter, self-hosting on GCP, AWS, or Azure, and a dedicated support contact in place of PromptLayer's own basic-tier support on every plan below it.
Four tiers, three meters
The four PromptLayer tiers, in plain numbers
Straight from PromptLayer's own pricing page, not a third party estimate.
Free: $0 a month
Five users, one workspace, ten saved prompts, ten playground runs a day. Twenty five hundred requests included, no listed overage rate past that, which reads as a hard cap rather than a billed one.
Pro: $49 a month
Same five user cap and starting allowances as Free, but unlimited workspaces, prompts, and playground runs. Overage past the included numbers bills at $0.003 per transaction instead of cutting off.
Team: $500 a month
Twenty five users, a hundred thousand or more requests, ten thousand or more agent node executions, webhooks included by default. A ten times price jump from Pro for a five times jump in seats.
Enterprise: custom pricing
No published number. Self-hosted on GCP, AWS, or Azure, or cloud-hosted single-tenant, plus RBAC, deployment approvals, SSO, and a signed HIPAA agreement, none included on any plan below it.
The number every tier bills against, undefined
What actually counts as one billable unit?
Every PromptLayer tier bills against three separate meters at once: requests, agent node executions, and eval cell executions. PromptLayer's own pricing page lists all three side by side on every row, and neither that page nor PromptLayer's documentation site defines what distinguishes one from the other in plain language anywhere public. The closest thing to an answer lives inside PromptLayer's own feature pages rather than its pricing page: its Workflows feature describes a visual Agent Builder for running multi-step agent workflows, which is presumably the source of an 'agent node execution', one step firing inside that builder, and its Evaluations feature describes graders running as columns against dataset rows inside a single evaluation table, which is presumably the source of an 'eval cell execution', one grader run against one row. PromptLayer never states either connection outright, so treat it as an informed reading of its own product pages rather than a confirmed definition.
That matters most once a workspace goes over its included allowance. Pro and Team both list an overage rate per 'transaction', $0.003 and $0.002 respectively, without ever saying whether a transaction means one request, one agent node execution, one eval cell execution, or some combination of the three counted together. A production agent with a five-step workflow and an evaluation attached to each response could plausibly trip all three meters on a single incoming message, which would make the real bill considerably higher than counting requests alone would suggest, though PromptLayer's own page gives no way to confirm that without running a workload and watching the invoice.
The one tier that changes where the data lives
Enterprise is the only tier with a hosting choice
Free, Pro, and Team all run as shared SaaS hosted in the US. Only Enterprise offers anywhere else.
Self-hosted on GCP, AWS, or Azure
PromptLayer's own infrastructure, deployed inside a customer's cloud account of choice, reachable only through the Enterprise tier and priced through a sales conversation rather than a published rate.
Cloud-hosted in the EU, or single-tenant
Two more Enterprise-only options: PromptLayer's standard cloud hosting moved to EU infrastructure, or a fully single-tenant deployment, neither available at any price on Free, Pro, or Team.
What the meter doesn't show
PromptLayer's meters tell you what the agent did. They tell your customer nothing
Every number on PromptLayer's pricing page, the request count against a plan's allowance, the agent node executions climbing inside a workflow run, the eval cell executions scoring a dataset, describes what happened inside your own account and what it cost your own team to run it. That holds at every tier, including Enterprise, since even its role based access control and deployment approvals still govern who inside your company can see or ship what, not whether an outside customer can see anything at all. Unlimited users on Pro and Team both mean unlimited seats for your own engineers, and PromptLayer's own multi-tenancy, one workspace holding many projects, still assumes every project belongs to your team, never to a customer sitting outside it. Even the self-hosted Enterprise deployment, running inside a customer's own GCP, AWS, or Azure account, changes where the data physically sits without changing who the dashboard was built for: the same engineers who had access on shared SaaS still have access on a single-tenant deployment, just on infrastructure their own company controls.
If you sell an AI agent inside a product other companies pay for, that gap shows up the moment a customer asks what the agent actually did for their account this month, a question no request count or eval score answers on its own. AiAgRe's Node SDK reads the same kind of agent events PromptLayer already logs, 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 PromptLayer's own prompt versioning or evaluation grading, and most teams run both: PromptLayer for what a prompt version did against a dataset, AiAgRe for what that same agent was worth to the person paying for it.
When a request count stops being the real question
Three questions PromptLayer's pricing page was never built to answer
None of these show up in a request count, a node execution, or an eval score. All three show up once real customers start asking.
Your customer asks what their account got, not how many nodes ran
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 workspace they can't log into. A passing evaluation reassures your engineers; it says nothing to the person deciding whether to keep paying.
Unlimited seats for your team, zero seats for theirs
Pro and Team both remove the meaningful limits as a workspace scales, more users, more workspaces, more prompts. Neither adds a scoped view for a customer outside your company, because PromptLayer's workspace model was built for your own projects, not for a second tenant layer underneath them.
The overage rate tracks volume. Nothing tracks value
The $0.003 to $0.002 per-transaction overage rate answers what an agent's traffic costs as it grows, on a unit PromptLayer's own page never fully defines. It says nothing about what that traffic was worth, the same gap a fuller look at AI agent evaluation tools as a category runs into past PromptLayer alone.
FAQs
PromptLayer pricing: frequently asked questions
Common questions from teams pricing PromptLayer against their own usage before they commit to a tier, or before they add a second layer on top.
How much does PromptLayer actually cost?
Four tiers, verified straight from PromptLayer's own pricing page. Free costs $0 a month for five users, one workspace, and ten saved prompts. Pro costs $49 a month, keeps the same five users but opens unlimited workspaces, unlimited prompts, and unlimited playground runs. Team costs $500 a month for twenty five users and a hundred thousand or more requests, the single biggest step in the whole table. Enterprise has no published number at all, quoted only through PromptLayer's sales team, and adds self-hosting, SSO, and a signed HIPAA agreement none of the three paid-by-the-month tiers include.
What's actually included on the Free plan, and how far does it really go?
Five users, one workspace, ten saved prompts, ten playground runs a day, and a ten megabyte cap on dataset storage, alongside twenty five hundred requests, seven hundred fifty agent node executions, and two hundred fifty eval cell executions a month, all support handled at PromptLayer's basic tier with no dedicated channel. The workspace cap is the limit that bites first for most teams, not the request count: a five-person team running two projects at once needs two workspaces before it needs the two thousand five hundredth request, and Free only ever grants one.
What is an 'agent node execution' and an 'eval cell execution', and how are they different from a request?
PromptLayer's own pricing page lists three separate meters, requests, agent node executions, and eval cell executions, side by side without ever defining any of the three in plain language, and neither PromptLayer's documentation site nor its product pages fill that gap either. The closest thing to an answer sits inside PromptLayer's own feature pages: its Workflows feature describes a visual Agent Builder for multi-step agent workflows, which is presumably what an 'agent node execution' counts, one step firing inside that visual builder, and its Evaluations feature describes graders running as columns against dataset rows in a single evaluation table, which is presumably what an 'eval cell execution' counts, one grader running once against one row. PromptLayer never states either connection outright, so treat this as an informed reading of its own product pages rather than a confirmed definition, and budget with the assumption that a single production request touching a multi-step agent with an evaluation attached could count against all three meters at once, not just the request meter.
What happens once a team goes over the Pro plan's request allowance?
PromptLayer's table lists Pro's requests, agent node executions, and eval cell executions with a plus sign after each number, twenty five hundred plus requests, seven hundred fifty plus agent node executions, two hundred fifty plus eval cell executions, and a flat overage rate of $0.003 per transaction next to that row. What counts as one transaction for that rate is never specified either: if it means one request, a Pro workspace running fifty thousand requests in a month would clear its included allowance and add roughly a hundred and forty two dollars in overage on top of the $49 base, close to $191 total, though that number is an illustration built on an assumption PromptLayer's own page leaves the reader to make, not a figure it states directly. Team's overage rate drops to $0.002 per transaction on the same undefined unit, and Free lists no overage rate at all, which reads as a hard cap rather than a billed one.
Why does the price jump from $49 to $500 with nothing in between?
Pro tops out at five users and the same twenty five hundred plus requests as Free before overage kicks in. Team jumps straight to twenty five users, a hundred thousand plus requests, ten thousand plus agent node executions, and its first included webhook support, all for $500 a month, a ten times jump in price for a five times jump in seats. Nothing sits between a five-person startup team and a twenty five person one on PromptLayer's own published tiers: a seven or eight person team outgrowing Pro's seat count has to either stay cramped on Pro and pay per-transaction overage, or absorb the full Team price meant for a team three times its size. The math usually favors staying on Pro longer than feels comfortable: even a heavy overage bill at $0.003 per transaction rarely closes a $451 gap to Team, so the seat cap, not the request meter, ends up forcing the upgrade decision for most growing teams.
Does PromptLayer Enterprise support self-hosting?
Yes, and it is the one capability that sets Enterprise apart from every other tier rather than just scaling the same numbers up. PromptLayer's page lists Enterprise as self-hosted on GCP, AWS, or Azure, cloud-hosted in the EU, or cloud-hosted single-tenant, a choice none of Free, Pro, or Team offer, all three running only as shared SaaS hosted in the US. Enterprise also adds role based access control, deployment approvals, and SSO, three features the pricing table marks as 'Not included' on every tier below it, plus a signed HIPAA business associate agreement and configurable data retention, none priced publicly and all reachable only by talking to PromptLayer's sales team. A team that needs SSO or RBAC alone, without the self-hosting or the HIPAA paperwork, has no smaller Enterprise-lite option on PromptLayer's own table; the only published path to either control is the full Enterprise conversation.
Is PromptLayer cheaper than LangSmith or Langfuse?
At the free tier, PromptLayer's five users and twenty five hundred requests sit well below Langfuse's fifty thousand units on its own free Hobby plan, though a PromptLayer request and a Langfuse unit measure different things and don't convert directly. Past free, PromptLayer's $49 Pro plan undercuts Langfuse's $199 Pro tier on sticker price alone, but Langfuse's Pro includes three year data retention where PromptLayer publishes no retention window on any tier at all. The fuller side by side, including where Langfuse and Confident AI fit into the same comparison, lives on the LangSmith alternatives page, and the full unit-by-unit Langfuse breakdown lives on the Langfuse pricing page.
Does any PromptLayer plan show pricing or ROI data to my own customers, and can I run AiAgRe alongside it?
No plan does, Free through Enterprise, and yes, running both is the normal setup rather than a workaround. Every dashboard PromptLayer renders, the request count climbing toward a plan's allowance, the evaluation table with its graded columns, the workflow's agent nodes mid-run, lives inside a workspace built for the team that owns it. Unlimited users on Pro and up means unlimited seats for your own engineers, not a scoped view for an outside customer, and even Enterprise's role based access control governs who inside your company sees what, never a second tenant layer underneath it. AiAgRe's Node SDK reads the same kind of underlying agent events PromptLayer already logs through its own tracing, tags each one with an organization identity and a customer identity at the point of ingestion, and turns that 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 PromptLayer counts a request or an agent node execution; 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 paying to run prompts and evals? Now show your customer what it did.
Request access and we'll walk through how AiAgRe's embed tokens map onto the trace data PromptLayer already produces.