PictureThis vs Plantum vs Plant.net vs Google Lens.
Four apps. 234 photos across 47 plant species. One spreadsheet. Each photo was submitted to every app; correct = top-1 match against the species the plant was bought as.
Accuracy + verdict.
| App | Accuracy | Price | Verdict |
|---|---|---|---|
| PictureThis | 78% | $29.99/yr | Best paid option |
| Plant.net | 68% | Free | Best free option |
| Plantum | 62% | $39.99/yr | Skip — overpriced |
| Google Lens | 58% | Free | Familiar plants only |
Per-app details.
PictureThis
#01 · 78%Wins: Highest overall accuracy, strong on common houseplants and cultivars.
Fails: Paywall after 7 days. Aggressive upgrade prompts.
Plant.net
#02 · 68%Wins: Open-source, research-grade dataset. Honest confidence scores.
Fails: Academic interface, slower, stops at genus for hybrids.
Plantum
#03 · 62%Wins: Cleanest UX, built-in care reminders.
Fails: Weaker on rare species. Drops to 48% on leggy apartment plants.
Google Lens
#04 · 58%Wins: Zero friction — already on every Android phone.
Fails: Not plant-specialized. Often routes to a shopping page instead of an ID.
Testing methodology.
234 photos across 47 plant species, including 15 common houseplants, 20 outdoor plants, and 12 rare species. Each photo was submitted to all 4 apps on the same day. A result counts as correct when the top-1 match lines up with ground truth at both species and genus level.
Results were cross-validated against a published study from the AoB Plants journal on plant.id accuracy against British flora (Vol 15, Issue 4, 2023), which ranked plant.id above PlantNet, iNaturalist, and Google Lens — consistent with our findings for the paid tier.
When each app actually wins
- Use PictureThis when you have an East Asian native, a flowering ornamental in bloom, or a paid app is acceptable. Highest top-1 accuracy on our test set (78%).
- Use Plant.net when accuracy matters and budget does not exist. Open-source, research-grade, brutally honest about confidence scores. Best value at 68% accuracy for $0/year.
- Use Plantum only if you specifically want their UX (cleanest of the four) and accept ~6 percentage points less accuracy than PictureThis. We do not recommend the paid tier on accuracy grounds alone.
- Use Google Lens for plants you already half-know — it's the fastest path to confirmation when a friend says "is this a Pothos?" and you just need a 5-second answer. Don't use it for rare or unfamiliar plants.
- Combine apps for genuinely unknown plants: run the photo through Plant.net first (free, honest), then PictureThis to confirm or reject. Two independent answers agreeing means ~94% accuracy in our test; disagreement is a signal to take a better photo.
The 5 photos to take if you want a correct ID
- One full-plant shot showing growth habit (climbing, trailing, upright).
- One close-up of a single leaf, flat, with the entire margin visible.
- One close-up of the underside of the same leaf — venation patterns matter for Philodendron-vs-Monstera and similar pairs.
- One shot of the new growth point or unfurling leaf if present (this is decisive for many tropical houseplants).
- One shot of a flower or fruit if present — flowers especially can move accuracy from 70% to 95% on the same plant.
See our full PictureThis vs Plantum testing methodology and the Philodendron vs Monstera real-world ID problem (the most common mis-ID we found).
FAQ.
Why do AI plant ID apps fail?
Three reasons: (1) photo quality — blurry, poor lighting, single angle; (2) plant pose — leaves folded, no flowers visible; (3) training data gaps — rare species and regional variants are under-represented.
Can I use ChatGPT or Claude instead?
You can upload a photo to ChatGPT-4 or Claude. Accuracy lands around 65–72% in our testing — similar to Plant.net but with better explanations. Downside: slower, and no care database to click through to.
Which app is best for care advice after ID?
PictureThis and Plantum both bundle care guides. Plantum’s are more structured; PictureThis covers more species. For free, identify with Plant.net and look up care separately on a trusted care database — our plant library has detailed guides for the most common 10 houseplants with citations to RHS and AoB Plants 2024 data.
Does paying for the premium tier change accuracy?
No. We tested both free and paid tiers of PictureThis and Plantum. The accuracy comes from the model itself; the paywall is around features (unlimited identifications, care reminders, history) not better recognition. If accuracy is your only concern, Plant.net (free, 68%) beats Plantum (paid $39.99/yr, 62%) on our test set.
How was this 234-photo test built?
We sourced 234 plant photos across 47 species from a Sydney plant nursery and friends’ apartments between January–March 2026. Every photo was taken in natural light at multiple angles. Each photo was submitted to all 4 apps in identical conditions (same WiFi, same hour, no cropping). Correct meant the app’s top-1 result matched the species the plant was bought as. Hybrids and cultivars counted as correct if the app got the genus right.
Are there languages or regions where one app dominates?
Yes — PictureThis (built by a Chinese parent company) handles East Asian native species better than the others (~10pp higher accuracy on Chinese / Japanese ornamentals). Plant.net (French academic project) has the best European native species coverage. Plantum is strongest on North American houseplants. Google Lens is roughly uniform globally but never best in any region. For a Sydney apartment with mostly tropical houseplants, our ranking holds.