`LanguageModelV3Middleware`

Language model middleware is an experimental feature.

Language model middleware provides a way to enhance the behavior of language models by intercepting and modifying the calls to the language model. It can be used to add features like guardrails, RAG, caching, and logging in a language model agnostic way.

See Language Model Middleware for more information.

Import

import { LanguageModelV3Middleware } from "ai"

API Signature

specificationVersion:

'v3'

The specification version of the middleware. Must be "v3".

transformParams?:

({ type: "generate" | "stream", params: LanguageModelV3CallOptions, model: LanguageModelV3 }) => PromiseLike<LanguageModelV3CallOptions>

Transforms the parameters before they are passed to the language model.

wrapGenerate?:

({ doGenerate: () => PromiseLike<LanguageModelV3GenerateResult>, doStream: () => PromiseLike<LanguageModelV3StreamResult>, params: LanguageModelV3CallOptions, model: LanguageModelV3 }) => PromiseLike<LanguageModelV3GenerateResult>

Wraps the generate operation of the language model. Receives both doGenerate and doStream functions.

wrapStream?:

Wraps the stream operation of the language model. Receives both doGenerate and doStream functions.

overrideProvider?:

(options: { model: LanguageModelV3 }) => string

Override the provider ID of the model.

overrideModelId?:

(options: { model: LanguageModelV3 }) => string

Override the model ID of the model.

overrideSupportedUrls?:

(options: { model: LanguageModelV3 }) => PromiseLike<Record<string, RegExp[]>> | Record<string, RegExp[]>

Override the supported URLs for the model.