Best AI for travel planning
You want a day-by-day plan for a trip, with real places, that fits your constraints. We gave nine assistants the same four-day Lisbon brief, no car and vegetarian, then checked every restaurant and every opening hour they came back with.
People also look for this as trip planning, itinerary planner, plan my trip or travel itinerary.
Our picks, side by side
ChatGPT
$8/mo · free tier: usableThe only plan usable without a follow-up. It put west and east Lisbon on separate days, cited opening hours and ticket prices that checked out, gave a budget, and pinned eight real vegetarian restaurants on a map. Its one error was a scheduling slip, lunch at a place that closes on Saturdays, not an invented place.
Perplexity
$20/mo · free tier: usableThe fullest plan of the nine, with three meals a day, a transport card, the weather, a vegan lunch in Sintra, and 35 citations. It also told us to book ahead at a Michelin-starred restaurant that closed in November 2025. Use it for the breadth, and check every restaurant before you book.
Gemini
$19.99/mo · free tier: usableEvery place it names is a live Google Maps card with today's open status, one tap from directions, which is the best format any tool produced for actually walking around with. The plan itself is too thin to use on its own, so get the plan elsewhere and have Gemini pin it.
1 of 8 named restaurants wrong (Manjerica scheduled on a Saturday; it is closed Saturdays). Castelo de São Jorge €15 and 09:00–21:00 verified; Pena Palace hours verified; Regaleira closing time was an hour late but last entry was right.
Wanderlog — A purpose-built trip planner with a map and a shared itinerary, which is what you end up wanting once a chat has given you the list.
Some links here earn us a commission and carry rel="sponsored". It never affects the ranking or the ratings — how we make money.
All 9 tools we tested
Rated for this task only · 0–3| Tool | Quality | Effort | Cost | Free tier | Measured | From |
|---|---|---|---|---|---|---|
| ChatGPTFree: Unlimited text chat on the default model, with limits on uploads, images, voice and deep research | 1/8 places wrong | $8/mo | ||||
| ClaudeFree: Chat with web search and file uploads, with a usage cap per five-hour session | 0/5 places wrong | $20/mo | ||||
| GeminiFree: The default model with usage limits, plus Gemini inside Maps, Gmail and Docs on a free Google account | 0/3 places wrong | $19.99/mo | ||||
| PerplexityFree: Unlimited basic searches with a weekly cap on Pro-model searches | 2/16 places wrong | $20/mo | ||||
| Microsoft CopilotFree: The Copilot app on web, Windows and mobile with usage limits | 1/7 places wrong | $19.99/mo | ||||
| DeepSeekFree: The full chat app with web search and file uploads; there is no paid consumer tier | 1/5 places wrong | Free | ||||
| GrokFree: Chat with web and X search within generous limits, no card required | 0/6 places wrong | $30/mo | ||||
| QwenFree: The full chat app; there is no paid consumer tier | 2/11 places wrong | Free | ||||
| KimiFree: Unlimited basic chat, file upload and web access on the free plan | 1/14 places wrong | $19/mo |
How we tested
One brief, sent once to each of ChatGPT, Claude, Gemini, Perplexity, Microsoft Copilot, DeepSeek, Grok, Qwen and Kimi on their free tiers, with no follow-up questions on the first pass. The brief asked for four days in Lisbon in mid-October for two adults on a mid-range budget, no car, one day trip outside the city, vegetarian, with specific places and a short list of restaurants.
Every restaurant each tool named was then checked against the venue's own site or listing for four things - does it exist, is it still open, is it genuinely vegetarian-friendly, and is it where the tool said it was - plus a fifth where the tool scheduled it, is it open that day. Quoted opening hours and ticket prices for the castle, Pena Palace and Quinta da Regaleira were checked against the venues. The count of wrong restaurants is the measured figure in the table.
All nine tools honoured all four constraints. Where they differed was in whether the plan could be used as written, and in what they got wrong. Five of them recommended something that would have gone wrong on the day - a restaurant that no longer exists, or a lunch at a place that is shut on the day it was scheduled. Kimi and Qwen, the two we ran as newcomers, produced the fullest plans of the group and each made exactly that kind of mistake.
The test was run by, or under the direction of, the one person who runs this site, with AI assistance, and every output and fact-check was reviewed by them. The surrounding text was drafted with AI assistance from our own ratings and evidence, then checked and edited by the same person. If something here is wrong, tell us and we will fix it and say so.
Questions people ask
Can an AI actually plan a whole trip for free?
Yes. Every tool here did the whole brief on its free tier in a single message, and ChatGPT's and Kimi's answers were usable as written. What none of them will do reliably is guarantee that a restaurant is still open, so budget ten minutes to check the places you care about before you book anything.
Which one hallucinated the least?
None of them invented a place that does not exist. The errors were subtler and more dangerous - restaurants that closed a year or two ago recommended as if they were open, lunches scheduled on the one day of the week the place is shut, a palace garden scheduled on a weekday when it only opens on Saturdays. Grok was the only one with nothing wrong, and it got there by not tying any restaurant to a day. Check days, not just names.
Why not a dedicated travel app instead?
If you want a map and a shared itinerary you can edit on the go, a purpose-built planner like Wanderlog is what you will end up in after the chat gives you the list. The assistants are better at the thinking part - what to do in what order and why - and worse at being something you carry.
Do I need the paid version of any of these?
Not for a four-day city trip. The free tiers produced everything above. The paid tiers buy you more messages and better models, which matters if you are planning something long or iterating a lot, not for a first draft.