Interface: EmbeddingInstance

@providerprotocol/ai

---

@providerprotocol/ai / @providerprotocol/ai / EmbeddingInstance

Interface: EmbeddingInstance<TParams>

Defined in: src/types/embedding.ts:190

Embedding instance returned by the embedding() function.

Example

const embedder = embedding({ model: openai('text-embedding-3-large') });

// Single input
const result = await embedder.embed('Hello world');

// Batch input
const batch = await embedder.embed(['doc1', 'doc2', 'doc3']);

// Large-scale with progress
const stream = embedder.embed(documents, { chunked: true });
for await (const progress of stream) {
  console.log(`${progress.percent}% complete`);
}

Type Parameters

TParams

TParams = unknown

Provider-specific parameter type

Properties

model

readonly model: BoundEmbeddingModel<TParams>

Defined in: src/types/embedding.ts:212

The bound embedding model

---

params

readonly params: TParams | undefined

Defined in: src/types/embedding.ts:215

Current parameters

Methods

embed()

Call Signature

embed(input, options?): Promise<EmbeddingResult>

Defined in: src/types/embedding.ts:198

Generate embeddings for one or more inputs.

Parameters

input

Single input or array of inputs

EmbeddingInput | EmbeddingInput[]

options?

EmbedOptions & object

Optional embed options

Returns

Promise<EmbeddingResult>

Promise or EmbeddingStream if chunked

Call Signature

embed(input, options): EmbeddingStream

Defined in: src/types/embedding.ts:202

Parameters

input

EmbeddingInput[]

options

EmbedOptions & object

Returns

EmbeddingStream

Call Signature

embed(input, options?): EmbeddingStream | Promise<EmbeddingResult>

Defined in: src/types/embedding.ts:206

Parameters

input

EmbeddingInput | EmbeddingInput[]

options?

EmbedOptions

Returns

EmbeddingStream | Promise<EmbeddingResult>