Embed API
Vector embeddings for brain search and similarity.
Requires AI provider (OpenAI, Anthropic, etc.)
Functions
| Function | What |
|---|---|
generate(text, options) |
Generate embedding vector |
generateBatch(texts[], options) |
Batch generate |
generateStack(text) / generateBatchStack(texts) |
Multibrain stack forms |
cosineSimilarity(a, b) |
Similarity score |
setProvider(name) |
Set active provider |
listProviders() |
List available |
getProviderInfo() |
Active provider info |
Usage
const embed = require('vant/lib/embed');
// Single embedding
const vec = await embed.generate('natural language query');
// Similarity (sync)
const sim = embed.cosineSimilarity(vec1, vec2);
// → 0.0-1.0 score
// Batch
const vecs = await embed.generateBatch(['query1', 'query2']);
Providers (hash, local, openai) are managed with setProvider /
listProviders - the CLI surface is vant embed set/list/info/generate
(see the CLI Reference).
Events
| Event | When |
|---|---|
embed:generating |
Before generation |
embed:generated |
After completion |
embed:batch:starting |
Before batch |
embed:batch:complete |
After batch |