PromptLayer Pricing: Plans, Costs & What You Get (2026)

PromptLayer pricing runs from $0 to $500 a month before Enterprise, and the number that sets your bill is five.

That's the user limit on Free and Pro, so a sixth person on the account moves you from $49 to $500. I priced five setups against today's rate card to show where each tier stops paying off.

TL;DR: PromptLayer Pricing

• PromptLayer costs $0 on Free, $49 a month on Pro, and $500 a month on Team, with custom Enterprise pricing.
• Free and Pro share the same allowance of 5 users and 2,500 requests a month. Pro's $49 buys unlimited prompts and workspaces, plus overage at $0.003 per transaction.
• Team raises the allowance to 25 users and 100,000 requests and cuts overage to $0.002. RBAC, SSO, a HIPAA BAA, and self-hosting all require Enterprise.
• A group of five or fewer stays cheaper on Pro until it passes 258,500 requests a month.

PromptLayer pricing plans at a glance

PromptLayer pricing uses four tiers with flat monthly fees. Each tier pairs a fixed user count with a monthly request allowance, and the two paid self-serve tiers bill extra usage per transaction.

Pricing verified against PromptLayer's pricing page on 23 September 2026. Verify with PromptLayer before you commit.

PromptLayer pricing plans in detail

Free plan at $0 per month

What's included: The Free plan gives you 5 users and 2,500 requests a month, plus 750 agent node executions and 250 eval cell executions. It also caps you at 10 prompts, one workspace, 10 playground runs a day, and 10MB per dataset.

Best for: Solo developers and small groups testing PromptLayer on a prototype before it sees customer traffic.

Pros:

✅ Five seats at $0, so a whole prototype group can share one prompt registry.

✅ Evals and agent workflows come included at small volumes, so you can try both before paying.

Cons:

❌ Ten prompts fill up fast once each feature gets its own template.

❌ The free plan lists no overage rate, so traffic past 2,500 requests means moving to Pro.

Pro plan at $49 per month

What's included: Pro keeps Free's 5 users and 2,500 requests, then lifts the caps on prompts, workspaces, and playground runs. Datasets grow to 150MB, and usage past the allowance bills at $0.003 per transaction.

Your $49 pays for unlimited prompts, workspaces, and playground runs, plus the option to keep going past the allowance at $0.003 per extra transaction.

Best for: Groups of up to five people running one or two LLM features in production at modest volume.

Pros:

✅ Overage scales in small steps, at $3 per 1,000 extra transactions.

✅ Unlimited workspaces, so staging and production prompts can live apart.

Cons:

❌ A hard limit of 5 users, with no add-on seats listed.

❌ Webhooks start on Team, so Pro can't push prompt events to your own systems.

Team plan at $500 per month

What's included: Team gives you 25 users and 100,000 requests a month, plus 10,000 agent node executions and 7,500 eval cell executions. It adds webhooks and 1GB datasets, and overage drops to $0.002 per transaction.

Best for: Groups of six to 25 people with production traffic above Pro's allowance.

Pros:

✅ Overage costs a third less than on Pro, at $2 per 1,000 extra transactions against $3.

✅ You get 40 times Pro's request allowance for about 10 times the price.

Cons:

❌ The $500 fee applies even to a six-person group sending 10,000 requests a month.

❌ RBAC and SSO wait for Enterprise.

Enterprise plan with custom pricing

What's included: Enterprise sets custom limits on every meter and adds role-based access control, deployment approvals, SSO, HIPAA with a BAA, dedicated support, and data retention control. Hosting options cover self-hosting on GCP, AWS, or Azure, cloud hosting in the EU, and single-tenant cloud.

Best for: Healthcare, finance, and EU-based companies that need a BAA, approval gates, or data kept in their own cloud.

Pros:

✅ The only tier with self-hosting and EU data residency.

✅ Deployment approvals let you require sign-off before a prompt version reaches production.

Cons:

❌ No published price. Budgeting starts with a sales call.

❌ Every tier below it runs cloud-hosted in the US only.

How PromptLayer billing works

PromptLayer bills a flat monthly platform fee plus per-transaction overage. Your user count decides which fee you pay, and your traffic decides how much overage lands on top.

Flat plan fees with fixed user limits

PromptLayer doesn't charge per seat. A five-person Pro account pays $49, the same as a solo developer on Pro. User six is the catch. Pro lists no add-on seats on the pricing page, so the next step up is Team at $500.

LangSmith bills the opposite way. Its Plus plan costs $39 per seat a month with unlimited seats, so eight people pay $312 before any trace usage. Per-seat pricing grows in even steps, unlike PromptLayer, which jumps at user six, then again at user 26.

Overage charges on requests, agent runs, and eval cells

PromptLayer meters requests, agent runs, and evaluation cell runs, and bills extra usage on each by transaction. Each meter carries its own monthly allowance, and overage only starts on the meter you exceed.

Watch the eval meter if you run regression evals on every prompt change. Pro includes 250 eval cell executions a month (30 times fewer than Team's 7,500).

The pricing page lists no log retention for Free, Pro, or Team. Only Enterprise mentions data retention control. Ask before you count on old request logs for audits or debugging, since a short window would erase last quarter's request history.

Billing runs monthly. There's no annual discount on the pricing page.

Security and hosting features reserved for Enterprise

PromptLayer holds SOC 2 Type 2, GDPR, HIPAA, and CCPA certifications according to its pricing FAQ. The BAA you need to send protected health information through it is an Enterprise feature, though.

Access controls also wait for Enterprise. If you need sign-off before a prompt version ships, or roles that separate editors from viewers, you'll need Enterprise's deployment approvals and RBAC.

PromptLayer cost examples for small and large groups

These five scenarios apply PromptLayer's published rates to illustrative usage profiles. I've assumed each execution past an allowance bills as one transaction, which matches how the pricing FAQ describes overage.

Three-person prototype on the Free plan

A three-person group sending 1,500 requests and 100 eval cells a month pays $0. The caps you'll feel first are the 10 prompts and the 10 playground runs a day, since both stop you once iteration picks up.

Five-person group at 40,000 requests a month

Picture five people sending 40,000 requests, 2,000 eval cells, and 1,000 agent node executions a month on Pro. The overage math adds up line by line.

• Requests: 40,000 minus 2,500 included equals 37,500 extra
• Eval cells: 2,000 minus 250 included equals 1,750 extra
• Agent node executions: 1,000 minus 750 included equals 250 extra
• Total: 39,500 extra transactions at $0.003 equals $118.50

Pro costs $167.50 a month for this group, which includes the $49 fee plus $118.50 in overage.

Sixth user and the jump to the Team plan

Take a lighter group of five sending 10,000 requests a month. On Pro, that's $49 plus 7,500 extra requests at $0.003, or $71.50 a month.

Add a sixth user and the bill becomes $500. Pro's 5-user limit forces the move to Team, a jump of about seven times for one extra login. Budget for Team the moment your hiring plan puts a sixth person on prompt work.

Pro versus Team crossover at 258,500 transactions

Pro stays cheaper than Team up to 258,500 transactions a month for a group of five or fewer, counting requests only. Pro costs $49 plus $0.003 per transaction past 2,500, and Team costs $500 plus $0.002 per transaction past 100,000. Both land at $817 at the crossover. Below that line, Pro wins.

At 150,000 requests, Pro costs $491.50 against Team's $600. The gap closes as traffic climbs. If your group also runs heavy evals or agent workflows, Team's larger eval and agent allowances pull the crossover lower.

Twenty-person group at 400,000 requests a month

A 20-person group sending 400,000 requests, 20,000 eval cells, and 30,000 agent node executions a month lands on Team. The overage covers 300,000 requests, 12,500 eval cells, and 20,000 agent node executions.

That's 332,500 extra transactions at $0.002 (about $2 per 1,000), or $665. Team costs $1,165 a month for this group. For scale, LangSmith Plus would cost $780 in seat fees alone for the same 20 people, before trace usage.

Choosing a PromptLayer plan by headcount and volume

Choose Free if you:

• Have up to five people exploring PromptLayer on a prototype.
• Send under 2,500 requests a month and keep fewer than 10 prompts.

Choose Pro if you:

• Have five or fewer people who edit or review prompts.
• Send up to about 258,500 requests a month.

Choose Team if you:

• Have six to 25 people who need their own logins.
• Need webhooks, 1GB datasets, or more than 250 eval cells a month.

Choose Enterprise if you:

• Handle protected health information and need a signed BAA.
• Need SSO, RBAC, deployment approvals, EU hosting, or self-hosting.

PromptLayer value by use case

PromptLayer pays off when people outside engineering edit prompts. Its prompt registry keeps templates outside your codebase, tracks version history, and releases a chosen version through labels.

PromptLayer is worth it if you:

• Have product managers or domain experts who change prompts every week and need a release process.
• Run a group of five or fewer at moderate volume, where Pro stays under $200 a month at 40,000 requests.

Skip PromptLayer if you:

• Need self-hosting or EU hosting without an Enterprise contract. Langfuse publishes a self-hosted option alongside its cloud plans.
• Have six to 10 people with light traffic. Team charges $500 for 100,000 requests you may never use, while Langfuse Core charges $29 a month with unlimited users.
• Ship voice or chat agents where the prompt is one variable among speech recognition, latency, and interruption handling. A prompt registry can't test those.

PromptLayer alternatives and pricing comparison

Five other tools compete for the same budget line, and each one bills on a different unit. Read every starting price next to its unit before you compare totals.

Free tiers vary as much as paid ones. LangSmith's Developer plan allows one seat and 5,000 base traces, Langfuse Hobby allows 2 users and 50,000 units, and Braintrust Starter allows unlimited users with 1GB of processed data.

For the full tier-by-tier math, see our Langfuse pricing and Braintrust pricing guides, or the Arize AI pricing guide if you bill by spans.

Platform fees for an eight-person group before usage

Eight people pay these platform or seat fees on each tool's cheapest plan that fits them, before any usage charges.

PromptLayer carries the highest fee in this table for an eight-person group. Its $500 does include 100,000 requests, so run your own traffic through each vendor's usage meter before you pick on seat fees alone.

Cekura vs PromptLayer for voice and chat AI agents

PromptLayer manages, versions, and scores prompts. Cekura tests the whole agent by running simulated multi-turn calls and chats against it before release, then scoring live production calls after launch.

Voice agents depend on more than prompt text. PromptLayer's LiveKit integration traces each STT, LLM, and TTS turn with its cost and latency. Tracing records calls that already happened. It won't put a simulated caller who interrupts mid-sentence in front of a new prompt version before release.

Cekura runs those tests at scale and publishes the results as public benchmarks.

Our voice AI benchmarks come from the same simulation engine: the agent-workflow benchmark put 8 voice platforms through 82 scenarios, where Retell passed 75.61% and LiveKit 70.73%.

A separate benchmark scored 270 live production calls across 9 platforms for voice quality, and the speech-to-speech benchmark ran 9 realtime models through 2,460 calls, with GPT Realtime 2.1 on top at 79.3%. Those are cross-platform, multi-turn numbers a prompt registry has no way to produce.

Cekura is better for:

• Running multi-turn simulated calls and chats against the full agent before each release.
• Scoring production calls for drop-off points, instruction following, and compliance checks.

PromptLayer is better for:

• Keeping one shared registry of prompt versions with release labels.
• Letting non-engineers edit and test prompts outside the codebase.

Use both if:

• Your prompt lives in PromptLayer and ships inside a voice or chat agent. Version the prompt there, then gate its release on an A/B test against the current version.
• You run a healthcare scheduling agent and need proof that each prompt edit still books, cancels, and reschedules correctly.

Real example: Quo, a Cekura customer, needed a way to "test every change to our voice agents with confidence," said Vanessa Cornelius, Prompt Engineer at Quo. Quo now runs regression and A/B tests on each prompt change before it ships.

Five features map to where a prompt change can go wrong:

Pre-production:

• A/B testing compares two agent versions side by side to measure a prompt, model, or configuration change.
• Chat testing runs text-based simulations at $0.025 per reply, a cheaper first pass before voice minutes.

Infrastructure:

The infrastructure suite ships pre-built scenarios for latency and stability across voice stacks.

Observability:

Production monitoring scores live calls at $0.05 per call.

Metric sampling evaluates a set percentage of calls, which trims evaluation cost on high-volume agents.

Native integrations cover Retell, Vapi, ElevenLabs, LiveKit, Pipecat, Bland, and more, so the agent you've already built stays where it is. Cekura supports SOC 2, HIPAA, and GDPR compliance, covering transcript redaction, role-based access, and audit trails.

Pricing starts with 300 free credits, about 60 minutes of voice testing, with no card required. The Startup plan costs $500 a month and includes 10 seats and about 2,000 voice testing minutes.

Book a demo to watch a simulation run against your own agent.

Bottom line on PromptLayer pricing

PromptLayer pricing is easy to predict until your sixth user. The flat fees hide nothing, the overage rates are public, and the only sharp step is the jump from Pro's 5 users to Team's $500.

Stay on Free until you pass 10 prompts or 2,500 requests. Move to Pro while five or fewer people touch prompts, and stay there up to about 258,500 requests a month. Plan for Team once a sixth person needs a login, and talk to sales about Enterprise only for a BAA, SSO, approvals, or self-hosting.

Pull last month's request, eval, and agent counts from your logs and run them through the formulas above before you pick a tier.

Frequently asked questions

How much does PromptLayer cost per month?

PromptLayer costs $0 on Free, $49 a month on Pro, and $500 a month on Team, with custom pricing on Enterprise. Pro bills $0.003 and Team bills $0.002 per transaction beyond each plan's allowance.

Is PromptLayer free?

Yes, PromptLayer has a free plan with 5 users, 2,500 requests, and 10 prompts a month on its pricing page. It lists no overage rate, so usage past those caps requires Pro or higher.

Does PromptLayer charge per seat?

No, PromptLayer charges a flat fee per plan with a fixed user limit. Free and Pro allow 5 users, Team allows 25, and Enterprise sets custom limits.

Can you self-host PromptLayer?

Yes, but self-hosting is Enterprise-only. Free, Pro, and Team run cloud-hosted in the US. Enterprise adds self-hosting on GCP, AWS, or Azure, EU hosting, and single-tenant cloud.

Is PromptLayer HIPAA compliant?

Yes, PromptLayer lists HIPAA compliance, but it signs a BAA for Enterprise customers only. It also lists SOC 2 Type 2, GDPR, and CCPA certifications.

What are the benefits of using PromptLayer?

The main benefits are a central prompt registry with version history and release labels, a playground for testing changes, request logging, and built-in evals. Prompts live outside your codebase, so non-engineers can edit and release a version by moving its label.

Which platform is best for prompt engineering?

The best platform for prompt engineering depends on who edits your prompts and where you host. PromptLayer suits groups where non-engineers edit prompts, LangSmith suits LangChain and LangGraph stacks, and Langfuse suits open-source and self-hosted setups.

What are the best prompt management and testing tools?

PromptLayer, Langfuse, LangSmith, and Braintrust cover prompt versioning and evals for LLM apps. Cekura covers multi-turn simulation and production call scoring for voice and chat agents.