by Stripe
Stripe's machine-learning fraud screen for online payments, trained on more than 70 trillion data points from the Stripe network, with optional abuse signals.
Use Radar if you already take payments on Stripe and have no fraud team: it works from the first charge and Stripe reports a 32% average cut in fraud. Pick a dedicated fraud platform instead if you process elsewhere, run a bank, or want chargebacks guaranteed rather than screened.
It is already on for every Stripe payment, scored with history from cards seen across millions of other Stripe businesses.
Lock-in comes from Stripe processing the payments Radar learns from; it weakens when a merchant routes volume through several processors.
Main trade-off
Its view is the payment and the Stripe account, not the full login and browsing journey that device and behaviour platforms watch.
| Plan | Monthly price | What it includes |
|---|---|---|
| Radar Lite | Free | AI fraud prevention for card payments, Card testing prevention, Fraud alerts, Included on Stripe standard payments pricing |
| Radar Standard | $10/mo | 200 screened transactions included, then $0.05 each, AI fraud prevention across all payment methods, Risk levels, default rules, allowlists and blocklists, Fraud analytics, fraudulent dispute and early fraud warning signals |
| Radar Plus | $14/mo | 200 screened transactions included, then $0.07 each, Risk scores 0 to 99, Custom rules with backtesting, manual reviews, risk preferences, Adaptive 3D Secure, Smart Refunds, Radar Assistant, agentic insights |
| Radar Pro | $20/mo | 200 transactions and 400 abuse events included, then $0.09 and $0.005, Dynamic risk thresholds, Free-trial, bot, multi-account and account-sharing signals, Pay-as-you-go abuse signal for usage-based billing |
| Enterprise | Custom | Custom pricing above 10,000 transactions, Radar signals for payments not processed on Stripe, Tailored recommendations |
Radar Lite is included at no extra charge on Stripe's standard payments pricing (card payments and card testing). Radar Standard $10 a month including 200 screened transactions, then $0.05 each, or $0.05 per screened transaction pay as you go. Radar Plus $14 a month including 200 transactions, then $0.07, or $0.07 pay as you go. Radar Pro $20 a month including 200 transactions and 400 abuse events, then $0.09 per transaction and $0.005 per abuse event. Platforms and marketplaces: Standard from $20, Plus from $44, Pro from $70 a month. Above 10,000 transactions, and for payments not processed on Stripe, pricing is by quote.
Stripe Radar is the machine-learning fraud screen inside Stripe that scores each payment, and on Pro, sign-up and free-trial abuse, using models Stripe says learn from more than 70 trillion data points across its network. It is turned on by default for Stripe merchants and adds rules, reviews and abuse signals on paid plans.
Network-trained risk scoring
Every payment gets a risk level on all plans and a 0 to 99 score from Plus up, from models trained on payments across the whole Stripe network.
Custom rules with backtesting
From Plus, write allow, block and review rules and test them against your own past payments before switching them on.
Radar Assistant
Describe a rule in plain language and Radar Assistant drafts it, so a founder without a fraud analyst can still tune the screen.
Adaptive 3D Secure
Plus and Pro ask the cardholder's bank for authentication only on payments Radar rates as risky, instead of on every checkout.
Abuse prevention signals
Pro adds signals for multi-account sign-ups, free-trial cycling, account sharing, bots and pay-as-you-go abuse in usage-based billing.
Connected-account risk for platforms
On Stripe Connect, Radar scores the merchants onboarded to a platform and can place reserves or restrict payouts on risky ones.
Radar outside Stripe
Enterprise customers can call Radar's payment, account and customer signals by API for payments that run on another processor.
The nearest AI tools on Hokai for the same job.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| $10/mo | Yes | Web | 3.9 | |
| -- | Trial | Web · iOS · Android | 4.1 | |
| Free | Yes | Web | 3.9 | |
| Quote | No | Web | 3.9 |
Proprietary
Radar Lite costs nothing extra on Stripe's standard payments pricing. Radar Standard is $10 a month with 200 screened transactions included and $0.05 for each one after, Radar Plus is $14 a month and $0.07, and Radar Pro is $20 a month and $0.09 plus $0.005 per abuse event beyond 400. Each plan can also be paid purely per transaction, and platforms pay from $20, $44 or $70 a month.
Yes, Radar Lite is included for businesses on Stripe's standard payments pricing and covers card payments, card testing and fraud alerts. It gives risk levels but no 0 to 99 score, no custom rules and no manual review queue, which start on Plus.
Lite protects card payments only and comes with standard Stripe pricing. Standard extends the screen to every payment method Stripe supports and adds fraud analytics, default allowlists and blocklists, and the fraudulent dispute and early fraud warning signals, per Stripe's plan comparison.
Yes, but only through Stripe's sales team. Stripe says Radar can be plugged into an existing setup without changing processor and its signals called by API to score payments, accounts and customers in your own system, priced by quote.
Only on Radar Pro. Pro adds a free-trial signal, a multi-account signal at sign-up, an account-sharing signal, bot limits and a signal for usage-based billing abuse, billed per abuse event after the 400 included each month.
Stripe Radar is the fraud prevention product built into Stripe, the payments company that says more than 5 million businesses build on it and that processed $1.9 trillion in payments volume in 2025. Radar scores every payment for fraud risk and can also score connected accounts on a platform and customers who abuse free trials, usage-based billing or sign-up flows. On a Stripe account it runs from the first charge with no setup, and Stripe now also sells its signals by API for payments processed elsewhere.
The mechanism is network scale: Radar's models train on payments across all of Stripe, and Stripe puts at 92% the chance a card paying a merchant has already been seen somewhere on Stripe. Each payment gets a risk level (normal, elevated, highest); paid plans add a 0 to 99 risk score, custom rules you can backtest against past payments, a manual review queue, adaptive 3D Secure that asks for bank authentication only on risky charges, and Radar Assistant, which writes rules from plain-language requests. Stripe reports that Radar users see a 32% average reduction in fraud, and that Anthropic cut wrongly blocked legitimate transactions by 83% with it.
It fits founders and small teams who sell online through Stripe Checkout, subscriptions or Stripe Connect and do not have a fraud analyst. The Pro plan is aimed at software and AI companies whose losses come from people opening many accounts, cycling free trials or running up compute before a card fails, a pattern Stripe calls token theft.
Radar is the wrong first pick when most of your volume runs through another processor and you need one engine across all of it, when you are a bank monitoring account-to-account transfers, or when you want a vendor to pay for fraud chargebacks it approved. Ecommerce guarantee vendors such as Signifyd take on that liability, and device and behaviour platforms such as SEON or Sift look further into the customer journey than a payment screen does. Compare the field on the AI security tools list.
Product facts on this page (pricing, plans, integrations, compliance) are read from Stripe's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.