Parezen is designed to cut your AI bill and prove every dollar against your own provider invoice. That proof is planned. Start free. Pay 15% of verified savings only if you choose.
Pricing updated 2 Oct 2026. Early access. The X-ray is planned and not live yet.
Free under $1K/month AI spend. No card, no licence key.
Read-only connectors are planned. No change to your traffic to start.
Designed to never block your traffic.
Plans that follow your AI spend
Parezen is designed as three modules: FinOps, Inference (the gateway) and Governance, on top of a free Core. Your plan band would be set by tracked monthly AI spend. Pick the modules you need, or take all three as the Suite and save 20%. Two modules save 10%.
Planned
Providers Parezen is built to work with, and would connect, are OpenAI, Anthropic, Google and Mistral. The gateway does not route to them today.
Your band: Starter
FinOps
$299
Inference
$99
Governance
$99
Suite
$399
Any two
$358 FinOps + Inference
Module prices by monthly AI spend. Prices are per month, billed annually.
Plan
Tracked AI spend / month
FinOps
Inference
Governance
Suite (all three)
Any two modules
Free
Under $1,000
$0
$0
Basic roles (in Core)
$0
n/a
Team
$1,000-$10,000
$49
$49
$29
$99
$88
Starter
$10,000-$30,000
$299
$99
$99
$399
$358 (10% off the two list prices)
Growth
$30,000-$100,000
$999
$500
$500
$1,599
$1,349 (10% off the two list prices)
Scale
$100,000-$250,000
$2,500
$1,000
$1,000
$3,599
$3,150 (10% off the two list prices)
Enterprise
Over $250,000, or SLA / custom terms
Custom
Custom
Custom
From $75,000 / year
Custom
Published terms: month-to-month billing is +20% over the annual price. This site does not take a payment.
Free
$0. Free under $1K/month AI spend.
No card. No licence key.
Organisation, workspaces, users and settings, on sample data
Spend view and request logs, on sample data
Planned
Provider connections and a usage export
Savings X-ray with read-only connectors
Savings simulator
Budgets with email alerts
Your own gateway (hybrid) and the open-source gateway
15% of verified savings, for 12 months per change, capped at 5% of the pre-change AI spend of the workloads we optimise (measured in your free X-ray). Needs paid FinOps. No platform fee on top. Verifying a saving against a provider bill is planned.
Planned
Free Savings X-ray
Free Savings X-ray: designed as an automated read-only look at your AI bill, for every plan. No card and no licence key. It is not a live connector.
$5,000
Verified Savings Assessment
Optional $5,000 Verified Savings Assessment, credited. It is designed to take 2 weeks, to be verified against the invoice, and to be credited in full against a contract signed within 60 days. Optional for Growth, Scale and Enterprise. It is required for Enterprise Offline.
Hybrid gateway licence
Planned
Hybrid is the default design. You run the gateway and its Postgres on your servers, next to your app. Parezen runs the dashboard, audit, ledger and billing. Your provider keys, prompts and responses stay in your environment. Parezen receives only signed, aggregated totals. Same modules, same prices, paid as an annual licence. Hybrid billing terms are proposed and may change before launch.
Free community licence. Designed never to block traffic.
Team
$588 / yr
$588 / yr
$348 / yr
$1,188 / yr
Starter
$3,588 / yr
$1,188 / yr
$1,188 / yr
$4,788 / yr
Growth
$11,988 / yr
$6,000 / yr
$6,000 / yr
$19,188 / yr
Scale
$30,000 / yr
$12,000 / yr
$12,000 / yr
$43,188 / yr
Enterprise hybrid
Minimum $90,000 / yr
The open-source gateway would be free with no licence key, at any spend. No image is published.
Savings share on top, the same on every deployment: 15% of verified savings, 12 months per change, capped at 5% of the pre-change AI spend of the workloads we optimise (measured in your free X-ray). It is designed to be verified from the signed totals your gateway sends (baseline vs actual by model and route, no prompts) and checked against your provider invoice through read-only billing connectors. Those connectors and that check are planned.
Growth and above would be able to buy through an AWS or Azure Marketplace private offer at the same price, and draw down eligible cloud commitments. Eligibility would depend on your agreement with AWS or Microsoft. Azure MACC does not count licences deployed on-premises or hybrid. Marketplace is not listed.
AWS and Azure Marketplace private offers are not listed. Contact sales.
Enterprise Offline (air-gapped): from $112,500 / year
Planned
For banks, government and anyone who allows no outbound traffic. Parezen would run fully offline. Images, routing-policy updates and price-book updates would arrive as signed offline bundles every month. Nothing would leave your network unless you chose to send it. Offline mode is not available.
Choose how your savings fee is set
Enterprise Offline savings fee options
(a) Flat annual fee
(b) Signed monthly report
15% of the estimated first-year savings, capped at 5% of the pre-change annual AI spend of the optimised workloads.
Your instance would produce a signed, tamper-evident monthly report: hash-chained totals, no prompts. You would export it and send it, with any provider invoice totals you choose to share.
An annual true-up would run against the signed year-end ledger if actual savings differ from the estimate by more than 10% either way.
We would bill 15% of the verified savings in the report (capped at 5% of the pre-change AI spend of the workloads we optimise (measured in your free X-ray)). Where you share invoice totals, we would bill the more conservative figure.
Audit rights once a year, with 30 days' notice.
A late report would have a 30-day catch-up window, then an estimate from the trailing 3 months, trued up when the report arrives.
The $5,000 assessment is required for Enterprise Offline. The offline X-ray would run locally in observe-only pass-through mode, so there would be no manual uploads.
Run it your way
Planned
Hybrid is the default: your gateway, our control plane. Hosted Quickstart lets a startup test by changing one URL. No install is published yet. Cross-provider switching, a smaller model, caching and prompt trimming are designs. The prices in this section are unchanged.
Hybrid: your gateway, our control plane (default)
Annual licence
The gateway container would run next to your app: Docker, Docker Compose, or a Railway or Render template in your own account. A Helm chart would follow. No published image.
It would use your own Postgres 15 or newer, or the bundled one. Your provider keys, prompts and request logs stay on your servers.
Only signed, aggregated totals would go up. A signed config would come down every minute, and budget hard stops would keep working if Parezen is unreachable.
The same OpenAI-compatible endpoint: change the base URL and use a pz_ key. Providers it would connect are OpenAI, Anthropic, Google and Mistral. It does not route to them today.
Hosted Quickstart: for startups
Proposed
Change one base URL to api.parezen.com and use a pz_ key. Nothing to run.
Bring your own provider keys. Parezen would hold them encrypted, and intake stays closed until encryption and rate limiting are on.
A free trial limited by time and volume, then a paid add-on priced above hybrid. The price is not set.
Metadata only by default. Request logs kept about 30 days. Move to hybrid at any time.
Hybrid via AWS and Azure Marketplace
Private offers
AWS would deploy on ECS or EKS. A private offer would be the purchase path.
Azure would deploy on AKS or Container Apps. A private offer would be the purchase path.
Same price as buying direct (Growth and above). It may count toward eligible cloud commitments.
GCP would deploy on GKE with the same Helm chart (direct purchase).
AWS and Azure Marketplace private offers are not listed. Contact sales.
Offline / air-gapped
Enterprise Offline, from $112.5K/yr
No outbound connection would be needed.
Signed offline bundles would arrive every month: images, routing policies and the price book.
The savings fee would be a flat annual licence or a signed monthly report. The install is not available.
Deployment comparison
Topic
Hybrid (default)
Hosted Quickstart
Offline / air-gapped
Where it runs
Gateway and Postgres on your servers; dashboard in Parezen's control plane
Parezen would run the gateway at api.parezen.com
Would run in your isolated network
What leaves your network
Signed aggregated totals only
Requests pass through Parezen; metadata only by default
Nothing would leave unless you send the monthly report
Provider keys
Stay in your gateway
Held by Parezen, encrypted
Stay in your network
Invoice check
Connectors would check the invoice
Connectors would check the invoice
Invoice totals you would choose to share
Updates
Signed images; prices and policies arrive in the signed config
Updates would be continuous
Signed offline bundles would arrive monthly
Pricing
Annual licence (12 × monthly); Enterprise from $90K/yr
Free trial, then priced above hybrid. Price not set
From $112.5K/yr
If the licence lapses
30-day grace, then pass-through
30-day grace, then pass-through
30-day grace, then pass-through
Only pay for savings the design can prove
The figures below are pricing examples. They are not savings measured on a customer bill.
Published pricing. The Verified Savings share is designed to charge 15% of verified savings, for 12 months per change, capped at 5% of the pre-change AI spend of the workloads we optimise (measured in your free X-ray). There's no platform fee on top. If we don't save you money, the share costs nothing. Verifying a saving against a provider bill is planned, not available.
If we don't save you more than we charge, you pay nothing that month. That sentence is the monthly savings share only. It does not apply to the subscription.
Illustrative example
A request used to cost $1.00. After Parezen's changes it costs $0.40, so you save $0.60. Our share is 15% of the saving ($0.09), but the fee can never be more than 5% of the pre-change spend: 5% of $1.00 is $0.05. So you pay $0.05 and keep $0.55 of the saving.
The cap only matters once savings pass about 33% of the original cost.
Illustrative example
Pre-change AI spend
Verified savings
15% share
5% cap (of $40,000)
You pay
$40,000
$8,000 (20%)
$1,200
$2,000
$1,200
$40,000
$16,000 (40%)
$2,400
$2,000
$2,000
How we calculate savings
Planned
These are the published rules. Measuring and reconciling them on a live bill is planned.
Your baseline is your own cost, not a list-price guess. An X-ray would measure what each request cost you before Parezen: your models, your caching, your batch share and your discounts.
The baseline would be re-priced at today's list prices. A provider price cut would go to you, and would not count as a Parezen saving.
Only a change Parezen would make would count, each for 12 months. Savings you already had, and anything you change yourself, would be excluded.
A request that fails your quality checks would be excluded. Model changes would go live only after shadow tests show no loss in quality.
The ledger would record baseline cost and actual cost by model and route. A self-hosted install would send signed totals only. An offline install would produce a signed monthly report.
Each month the ledger would be reconciled with your provider's bill, and the more conservative figure would be billed.
The formula on this page is published. Your ledger and price book would be open to export, so you could recompute every number.
Over $250K a month, or anyone who needs an SLA or custom terms. Enterprise starts at $75K a year hosted, $90K self-hosted and $112.5K offline. The intended purchase paths are direct or an AWS or Azure Marketplace private offer. Marketplace listing is not live. Hosted, self-hosted and offline installs are planned.
Published definition. Your plan band is your monthly AI bill. Providers Parezen is built to work with, and would connect, are OpenAI, Anthropic, Google and Mistral. That connection is not live, and the gateway does not route to them today.
No. Those are the published terms. The Free plan needs neither. The open-source gateway is designed to be free with no key at any spend, and a gateway you run (hybrid) is designed to get a free community licence under $1K a month. No gateway image is published, and Hosted Quickstart is not available.
Planned. Your traffic is designed to keep flowing. You get 30 days of grace with alerts. After that, Parezen is designed to switch to pass-through: requests are forwarded unchanged, while optimisation and paid features pause until you renew. Data export is designed to keep working. This is not live.
No. That is the design, and it is planned: billing never causes an error on your production traffic. Your gateway runs next to your app, so if Parezen's control plane is unreachable it keeps serving on its last signed config. Only the budget hard stops you set yourself return an error. There is no live gateway.
Published pricing rule, not a live bill. 15% of verified savings for 12 months after each change, capped at 5% of the pre-change AI spend of the workloads we optimise (measured in your free X-ray). Savings are your pre-Parezen cost, re-priced at today's list prices, minus your actual cost. Only a change Parezen would make counts, requests that fail quality checks are excluded, and the more conservative of the ledger and your invoice would be billed. Checking a line against a provider bill is planned.
No. The published rule re-prices your baseline at today's list prices, so price cuts go to you, not into our fee. That re-price is not running on a live bill.
No. It's optional, and it needs a paid FinOps module. You can use Parezen on subscription only.
Planned. The intended install is Docker, Kubernetes (Helm), AWS (ECS/EKS) or Azure (AKS/Container Apps), or a fully offline install. None of those images are published. Self-hosted plans are billed as an annual licence equal to 12 × the monthly list price, paid upfront. Enterprise self-hosted starts at $90,000 a year.
Planned. On a connected self-hosted install, only signed, aggregated totals (baseline vs actual cost by model and route) are designed to leave your network, never prompts or responses. On an offline install, nothing is designed to leave unless you send the monthly report. Neither install exists yet.
Planned. Images, routing policies and price-book updates are designed to arrive as signed offline bundles every month. For the savings fee, the choice is either (a) a flat annual fee set from your $5,000 assessment, with a yearly true-up and audit, or (b) a signed monthly ledger report that you export and send. Enterprise Offline starts at $112,500 a year. Offline mode is not available.
Not available today. The intended path is a private offer, at the same price as buying direct, for Growth and above. It may count toward your AWS or Azure commitment, depending on your agreement. Azure MACC does not cover licences deployed on-premises or hybrid. Marketplace is not listed.
Planned. The free Savings X-ray is designed as an automated read-only look at your AI bill. It is the intended way in for every plan, with no card and no licence key. Optional $5,000 Verified Savings Assessment, credited. It is designed to take 2 weeks, to be verified against the invoice, and to be credited in full against a contract signed within 60 days. Optional for Growth, Scale and Enterprise. It is required for Enterprise Offline.
Published terms, not a checkout. Yes, on hosted plans, at +20% over the annual price. Self-hosted licences are annual only. This site does not take a payment.
Published rule. Bands use your trailing 3-month average. You move up only after two months above a limit, and move down at renewal. Live billing that applies this rule is not built.
No. Parezen is designed to read LiteLLM, OpenRouter and Portkey logs and your provider bills, and to prove savings on top of the gateway you already run. That log import is planned.
Yes. 50% off Team and Starter for your first year. Limited places. Apply and we will confirm whether a place is open.