Two models, one trip format
MarcoPolo, trained from scratch on travel, plans from our catalogue of real places. MarcoPolo Navigator plans with maps, places and live data, in the voice you give it. Both return the same trip format.
Request
curl https://api.tripcreator.ai/v1/developer/trips \ -H "Authorization: Bearer $TRIPCREATOR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "marcopolo", "prompt": "A week in Italy, north to south, art and coast.", "country": "Italy", "days": 7 }'
Response
200 OK
{
"id": "trp_7Hq2",
"model": "marcopolo",
"environment": "sandbox",
"trip": {
"title": "Art and Coast Across Italy",
"segments": [
{
"location": {
"city": "Florence",
"country": "Italy"
},
"days": [
{
"day_number": 1,
"day_title": "Florentine Markets and Historic Stalls",
"activities": [
{
"sequence": 1,
"name": "Florence Central Market",
"entity_id": "3458396",
"latitude": 43.7773,
"longitude": 11.2538
},
{
"sequence": 2,
"name": "Mercato del Porcellino",
"entity_id": "3463217",
"latitude": 43.77,
"longitude": 11.2543
},
…
]
},
…
]
}
]
}
}The same 8 trip briefs, planned by MarcoPolo and by GPT-6, Gemini 3.8 Flash and Claude Opus 4.8. MarcoPolo was up to 4.6x faster and 10.8x cheaper per trip day.
See the full benchmarkThe general models invented one in 701, checked by hand.
Each one is pinned on the map, ready for your team to quote. None of the general models' stops are.
Median in MarcoPolo Chat on production. The first day is on the map in about 5 seconds.
A general model costs $0.17 or more a day once each stop is checked.