What “semantic search” means
Semantic search understands what you mean, not just what you typed. Describe a scene in a full sentence — the light, the emotion, the moment — and Pexafy ranks photos by how closely they capture that idea. No tag-guessing, no boolean operators, no scrolling through pages of near-misses. The richer your description, the better the match.Three ways to search
1 · Text search
Describe a scene in plain language and get photos that match its meaning.
2 · Image search
Upload a photo and find visually similar images — semantic reverse image search.
3 · Image + text
Start from a picture, then steer the results with words.
1. Text search — search images by description
The fastest way in: type what you picture, in plain English. Because the search is semantic, full descriptive sentences beat bare keywords. Here is a real Pexafy search for a query no keyword index would handle well:
2. Image search — turn a watermarked thumbnail into free photos
This is where semantic reverse image search earns its keep. Say you found the perfect shot on a paid stock site — but it’s a low-resolution, watermarked preview you’d have to license. Drop that thumbnail into Pexafy and it finds free, royalty-free photos of the same scene — same subject, mood and composition — that you can actually use. Here’s a watermarked stock thumbnail of a group of friends:

3. Image + text — start from a picture, steer it with words
The most powerful mode combines both: give Pexafy a reference image and a few words, and it balances what the picture looks like with what you describe. Take the same group photo and add a direction like “with hands in the air” — and the results shift toward that moment while keeping the original look and feel:
Start building
Get an API key
Run your first semantic search in minutes.
Use Pexafy in Claude
Search free stock photos right inside your AI assistant.