POST
/api/v1/engine/valuationread:engineAutomated valuation from an address or GPS point, the floor area and the property type. A flat returns the market value with its Q25-Q75 band, the asking price and the statutory values; a house returns the band and the statutory values only, no point price. The Flat / House toggle shows each shape.
| Name | Type | Example | Description | |
|---|---|---|---|---|
country | enum | required | sk | your two-letter country code (sk | cz | pl | si | hu | us). The response echoes the detected country. |
address | string | Location option A | Mýtna 21, 811 07 Bratislavamytna 21 bratislava | location requirement, option A: free-form address: street with number and a city, optionally a postal code, in any order, diacritics optional. Ambiguous street name → add a city or postal code. |
lat + lon | number | Location option B | 48.1512, 17.1083 | location requirement, option B: GPS coordinates (decimal degrees WGS84) instead of the address; the building is taken at the point. |
usable_area | number | required | 65 | net usable living area in m², [5, 10000] |
property_type | enum | required | flat | flat | house |
disposition | enum | 2_room | studio | 1_room | 2_room | 3_room | 4_room | 5_plus_rooms. The room count; for an atypical unit (loft, maisonette), send its room count. | |
condition | enum | new_build | original_state | partial_renovation | full_renovation | new_build. The renovation ladder, priced strictly monotone in this order; new_build is priced on the dedicated new-build model. | |
construction | enum | brick | brick | panel | reinforced_concrete | mixed | skeleton | wooden | stone | prefab | |
ownership | enum | private | private | cooperative | company | state | other | |
energy_rating | enum | B | A | B | C | D | E | F | G | |
furnished | enum | furnished | furnished | partially_furnished | unfurnished | shell | |
floor | number | 3 | [-3, 100]; 0 = ground | |
total_floors | number | 6 | [1, 100] | |
floor_area | number | 65 | total floor area in m² (incl. internal walls); defaults to usable_area when omitted | |
land_area | number | 450 | m² (for houses) | |
year_of_construction | number | 2022 | [1800, 2100]; enables the technical value (depreciation) | |
renovation_year | number | 2015 | [1800, 2100] | |
bathroom_count | number | 1 | [0, 20] |
curl -X POST https://api.bytero.ai/api/v1/engine/valuation \
-H "Authorization: Bearer bk_live_..." \
-H "Content-Type: application/json" \
-d '{
"country": "sk",
"address": "Mýtna 21, 811 07 Bratislava",
"usable_area": 65,
"property_type": "flat",
"disposition": "2_room",
"condition": "new_build",
"construction": "brick",
"ownership": "private",
"furnished": "furnished",
"energy_rating": "B",
"floor": 3,
"total_floors": 6,
"year_of_construction": 2022
}'| Field | Description |
|---|---|
country | the detected country of the property |
market_value_eur | the market (sale) price, EUR. Returned for flats only; a house is valued to a band, so this field is omitted and price_band is the headline |
price_band | the Q25-Q75 price band around the value ({ q25, q75 }, EUR), from the model's own quartiles; the value always falls inside it. For a house this band is the headline (no point market_value_eur) |
predicted_asking_eur | the asking price the same property would list at, EUR (flats only) |
confidence | high (20+ comparable listings near the address) | medium (10+) | low (fewer); degraded answers never claim high |
statutory_values | the vyhláška 492/2004 figures: insurance_sum_eur (reproduction value, the basis of the insured sum) always when the area resolves; technical_value_eur (reduced for age and wear) when year_of_construction is given |
model | which model produced the number: branch (flat / new-build / house), version, how the price band was derived, and whether the asking-to-sale adjustment applied. basis is "model" normally, "comparables-fallback" when the model was briefly unavailable (never claimed as high confidence) |
median_price_per_m2 / comparables_band_eur / comparables_used | context from comparable listings near the address; comparables_band_eur is their { q25, q75 } in EUR |
ml_location_used | the location context the model actually used (city, district, region), derived server-side from the address |
address_match | how the address resolved: matched address, coordinates, precision, match score, and alternatives when the address exists in several towns |
inputs | echo of the normalized payload that was priced |
warnings | present only when something needs attention: ambiguous address, model fallback |