Thanks. Open the lab to check your subscription status.
Test and train shopping agents at real checkouts
Pay per checkout request on the API, or reserve private environments built from your store.
The testing lab and an API sandbox key for teams evaluating shopping agents.
Includes:
- Testing lab access
Checkout Bench API sandbox key- Agent guide and store notes
- Public leaderboard and research
- 3 connected store accounts
- 5 concurrent agent runs
Environments
- Top 10 U.S. retailers (The CB10 Index)
- Run replays for 7 days
The
Checkout Bench API for teams shipping shopping agents, with usage that scales as you grow.
Includes:
- Everything in Lab
- Usage priced per checkout request as you grow
- Live checkout requests at real stores
- Pay by card
- Standard email support
- 25 connected store accounts
- 25 concurrent agent runs
Environments
- Every store in the index
- Full traces and replays for 90 days
Private training environments and data built from your own store, for retailers and model labs.
8 slots open for December 2026Includes:
- Everything in Developer
- Zero data retention
- Private environments for your store
Available on custom plans
- Custom MSA and DPA
- SSO and SCIM for the lab
- Monthly invoicing
- Dedicated Slack channel
- Enterprise SLAs
- Custom concurrency
- Training data exports
- Bring your own model or gateway
Products
- Shopper request in, checked cart and total out
- Signed in as the member at real stores
- Ready for the person to approve
- Any model through a real checkout
- Every step replayed and graded in code
- Model and compute included
- Weekly agent checks on your store
- Where agents stall, step by step
- Side by side with leading retailers
- Training environments from your checkout
- Replayable for training and evaluation
- New journeys added every week
Add private environments with an Enterprise plan
Keep your store's journeys, traces and training data to yourself, with zero retention of requests and results on supported products.
- Zero data retention on supported products
- $30,000 annual commitment
- Private environments for your store
Use cases
Shopping assistants
Let your assistant buy from real stores, with a checked cart and total the person approves.
- Send a shopper's request and get back a ready checkout
- Signed in as the member, with their address and saved card
- Questions come back when the store needs an answer
Retailers
See how AI agents shop your store and fix where they stall before your customers' agents arrive.
- Weekly scorecards against leading retailers
- Every stall traced to the page and step
- Training environments built from your checkout
Model labs
Train and evaluate shopping agents on real, replayable checkouts instead of scraped pages.
- Environments built from signed-in journeys
- Graded in code, never by another model
- Exports for training and evaluation
Payments
Test how agent payments and wallets hold up at real checkouts before you ship them.
- Link, Plaid and card flows at real stores
- Every total checked line by line
- Live orders only with a person's explicit yes
Endpoint pricing
| Checkout request | Answer and resume | Lab run | Trace export | Scorecard refresh | |
|---|---|---|---|---|---|
| Base price | $0.25 | $0.05 | $0.40 | $0.01 | $2.00 |
| Each extra item in the cart | $0.05 | – | $0.05 | – | – |
| Frontier model surcharge(per run) | – | – | up to $1.60 | – | – |
| Unit | per request | per answer | per run | per run | per store |
Partner credentials
Sign agents in and pay at checkout through the partners you already use, with no separate contracts for self-serve providers.
See integrationsSelf-serve
- Nook
- Link by Stripe
- Privy
Available on request
- Plaid
- Shopify merchants
Benefits
- Agents shop signed in as the member
- Bring your own model or gateway key
- Keys stay on our side, never with the agent
Lab and Developer are billed monthly by card. Usage prices are early-access rates that may change and are not charged automatically. Taxes may apply at checkout.