Best AI for gardening
You want a plant identified from a photo, told what is wrong with it, and told what to do this week. We sent two photographs with a known plant and a known problem to nine assistants and checked every answer against the truth.
People also look for this as plants, plant identification, plant care or what is wrong with my plant.
Our picks, side by side
ChatGPT
$8/mo · free tier: usableBoth plants, both problems, and the plan a gardener would write - alcohol on a cotton bud today, soap tomorrow, look inside the crown, come back in a week because the eggs survive; prune the mildewed begonia to the bin, keep the rest, don't repot a stressed plant, don't feed it this week. It is the one answer that explains that powdery mildew is about wet leaves and still air rather than wet soil, and it gives you the sentence to say at a Portuguese garden centre.
Grok
$30/mo · free tier: usableThe only tool to name both organisms exactly - citrus mealybug, Planococcus citri, and the begonia mildew, Oidium begoniae - which is what you type into a search or say at the counter. Every step in both plans is sound, down to checking the pot rim and drainage holes and not composting the cuttings. Six searches per photo, in a private chat.
Microsoft Copilot
$19.99/mo · free tier: usableBoth right, and it tells you its confidence - "less certain of the plant ID than of the pest" on the Clivia, "90% likely" on the begonia - which is exactly what you want before you spray anything. The plans are today / tomorrow / next week and cover the crown and the pot rim. Attaching the first photo opens a one-time consent dialog about people in images; approve it once.
2 of 2 plants identified (Clivia miniata; bronze-leaved wax begonia), 2 of 2 problems diagnosed (mealybugs; powdery mildew), nothing harmful in either plan, the watering and the north terrace addressed in both.
Pl@ntNet — Three of the nine assistants named the wrong plant from a clear photo. If the name is all you need, a dedicated identification app compares your photo against millions of tagged ones and shows you the matches to judge yourself; then ask an assistant what to do about it.
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 | 2/2 identified · 2/2 diagnosed | $8/mo | ||||
| ClaudeFree: Chat with web search and file uploads, with a usage cap per five-hour session | 2/2 identified · 2/2 diagnosed | $20/mo | ||||
| GeminiFree: The default model with usage limits, plus Gemini inside Maps, Gmail and Docs on a free Google account | 2/2 identified · 2/2 diagnosed | $19.99/mo | ||||
| PerplexityFree: Unlimited basic searches with a weekly cap on Pro-model searches | 1/2 identified · 2/2 diagnosed | $20/mo | ||||
| Microsoft CopilotFree: The Copilot app on web, Windows and mobile with usage limits | 2/2 identified · 2/2 diagnosed | $19.99/mo | ||||
| DeepSeekFree: The full chat app with web search and file uploads; there is no paid consumer tier | 1/2 identified · a mite killer prescribed | Free | ||||
| GrokFree: Chat with web and X search within generous limits, no card required | 2/2 identified · both species named | $30/mo | ||||
| QwenFree: The full chat app; there is no paid consumer tier | 2/2 identified · 2/2 diagnosed | Free | ||||
| KimiFree: Unlimited basic chat, file upload and web access on the free plan | 1/2 identified · full sun for a Clivia | $19/mo |
How we tested
Two openly licensed photographs whose plant and problem were identified by the photographers - a Clivia with citrus mealybug, and a bronze-leaved wax begonia with powdery mildew, the second shot by a plant pathologist. Each went to ChatGPT, Claude, Gemini, Perplexity, Microsoft Copilot, DeepSeek, Grok, Qwen and Kimi on their free tiers with one line of context - a pot on a north-facing terrace in Lisbon, watered every three days - and the question what plant is this, what is wrong, what should I do this week. A new chat per photo, no follow-ups.
We checked the plant name, the diagnosis, whether the plan fits the problem, whether the answer used the light and the watering it was given, and whether anything in it would hurt the plant. For the Clivia, a good answer says every three days is too wet. For the begonia, a good answer keeps water off the leaves and does not blame the schedule.
All nine named both problems - the white fluff and the white powder are hard to miss. The plants were harder. Six tools named both. Perplexity read the Clivia as a Dracaena, DeepSeek as a Yucca, Kimi as an agave - and Kimi, believing it had an agave, told us to move the plant into full sun and repot it in cactus mix, which would scorch a Clivia. DeepSeek prescribed "Acaricide" with imidacloprid for the mealybugs, a mite killer by name and an active ingredient not sold to amateur gardeners in the EU. Grok was the only tool to name both organisms to species. Kimi answered the begonia in Russian; Gemini's begonia answer arrived under a medical disclaimer.
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 identify a plant from a photo?
Usually, from a clear photo. Six of nine named both plants here, most to species. But three read a Clivia - broad strap leaves, hard to mistake in person - as a Dracaena, a Yucca and an agave, and one of those then gave advice that fits an agave and would damage a Clivia. If the name matters, get it from two sources, or use a dedicated identification app and ask the assistant what to do next.
Which tool gave the safest advice?
Seven of the nine gave plans with nothing harmful in them. The two that did not both got the plant wrong first. The lesson is that the diagnosis was easy and the identification was where the risk sat, so check the name before you follow the plan.
Did they use the context I gave?
All nine read the north-facing terrace and the three-day watering, and most told the Clivia owner to water less, which is right. On the begonia, the better answers - ChatGPT, Claude, Grok - said the schedule was not the cause and the wet foliage and still air were; the weaker ones blamed the watering anyway.
Do I need to attach a photo?
Yes, for identification. Every tool here accepted an image on its free tier, including DeepSeek, which our plan expected to read text only. Copilot asks you to confirm once that any people in your images have agreed to it; a plant photo still gets the dialog.