Search
Semantic, deep and multimodal scene search.
Semantic search
Search the indexed video corpus by natural-language query. Returns scene-level matches grouped by parent video, scored across caption and optionally transcript and multimodal embeddings.
Multimodal search (image / audio query)
Same as `/v1/search` but the query is an **uploaded image or audio file** instead of text — useful when text is a poor description of what you're looking for. Multimodal embeddings on the query side; results come back in the `omni` column only.
Deep search (multimodal rerank)
Like `/v1/search` but applies a second-pass multimodal rerank over the top scenes. Higher quality at the cost of latency (typically +2–5s).