It shows its working
Every result carries the applied policies, the income lines after shading, the buffered commitments and the reason for each adjustment — in plain English, not just a total.
Model Context Protocol · built for agents
Serviceability for free.
Ten Australian lenders, one client scenario, one call — borrowing capacity, credit policy and carded rates, with every number showing its working.
- 10 lenders modelled
- 8 tools
- No UI to click through
- Estimates, always labelled
> Paramedic on $125k + $18k overtime, partner on $78k, one kid, $15k card, $25k HECS. Buying at $1.1m in NSW. Who lends the most? ⏺ cheapli · run_serviceability applicants: 2 dependants: 1 state: "NSW" amount: omitted — solve each lender's maximum pass CBA $1,000,797 9.14% DTI 4.6 pass ING $993,763 9.04% DTI 4.6 pass ANZ $951,044 9.54% DTI 4.4 …7 more spread $49,753 across the panel. Drivers: ANZ assesses 0.40% above the median rate; the paramedic's overtime counts at 100% under CBA's essential-services rule.
$0
per month · per seat · per scenario
The pricing page is this paragraph.
No trial that expires, no per-user tier, no "contact us for enterprise". It is an estimate engine built on public data — charging you for arithmetic felt rude.
Fair use applies: the browser demo is rate limited, and connected agents get sensible per-token limits so one enthusiastic loop can't take the service down for everyone else.
Three reasons it's worth what you're not paying
Cheapli isn't a calculator with an API bolted on. It's a modelled serviceability engine that hands an agent the numbers and the reasoning.
Ten lenders, one call
One scenario in, the whole modelled panel out — sorted by capacity, with a summary of the spread and the specific drivers that put one lender ahead of another.
Made for agents, not tabs
No dashboard, no logins for your team, no screen to learn. It speaks MCP, so your assistant discovers the tools, calls them, and answers in your words.
Three steps, about four minutes
Connect once
Add the endpoint to Claude Code, Claude Desktop or your own agent. One command, one token.
Ask in plain English
"Can these two service $850k?" · "Which lender treats casual income best?" · "What's the duty on $1.1m in Victoria for a first-home buyer?"
Get numbers with provenance
Capacity, surplus, DTI and status per lender — each stamped with the model version and what it was last checked against.
Eight tools your agent can reach for
Plus resources: each lender's policy notes, the state duty schedules and the household expense benchmark table.
run_serviceability
One scenario against every modelled lender — capacity or an assessment of a requested amount, with the full breakdown.
get_policy
Citable policy entries by topic: casual income, HECS, credit cards, self-employed, notional rent, DTI, LVR and more.
search_products
Current carded rates by lender, purpose, repayment type, fixed term and LVR tier — refreshed from public open-banking data.
list_lenders
What's modelled, at which model version, verified against what, and how fresh each lender's rate data is.
calc_stamp_duty
Transfer duty and government fees for all eight states and territories, including first-home-buyer concessions.
calc_lmi
Indicative mortgage insurance premium, duty on the premium, and what capitalising it does to the loan.
calc_repayments
P&I or interest-only repayments at any frequency, plus total interest over the term.
calc_funds_to_complete
The whole settlement position itemised: duty, fees, insurance and legals against the deposit.
Lenders modelled
Ten of the largest Australian home-loan books. Each is a hand-built, version-controlled model of that lender's published serviceability approach.
Commonwealth BankWestpacNABANZMacquarie Bank
INGSuncorp BankBankwestSt.George BankAMP Bank
Named for identification only. Cheapli is independent — not affiliated with, endorsed by, or supplied data by any of them.
Where the numbers come from
Serviceability models are built by hand from public lender calculators, published broker policy material and the regulator's buffer guidance. Each one is a reviewed change in version control — never auto-scraped into production.
Rates come from the banks' public Consumer Data Right product APIs, refreshed daily. They're advertised carded rates: negotiated discounts aren't in there.
Policy answers are derived from the same configuration the engine executes, so what Cheapli says a lender does can't drift from what it calculates.
Honest limitations
- Every figure is a modelled estimate, not a lender decision. Real assessments differ.
- Household expense benchmarks are a labelled public approximation of the licensed industry benchmark.
- Mortgage insurance premiums are indicative, from a public premium grid.
- Carded rates only — no discretionary pricing.
- Models carry a version and a "last verified against" stamp; check them before you rely on them.
Point your agent at it
Run a scenario in the browser right now, or request a token and have it answering questions in your terminal by lunchtime.