`@anvia/mistral`

Mistral’s provider adapter supplies completion, embeddings, OCR, and model listing. It is useful when one server workflow needs both generative models and document extraction without changing Anvia’s core contracts.


Support	First-party
Version	`0.4.1`
Runtime	ESM, server-side JavaScript
Peer	`@anvia/core >=0.7.1 <1.0.0`

Install

bash

pnpm add @anvia/mistral @anvia/core

Create a completion model

import { AgentBuilder } from '@anvia/core'
import { MistralClient } from '@anvia/mistral'

const mistral = new MistralClient({
  apiKey: process.env.MISTRAL_API_KEY,
})

const agent = new AgentBuilder(
  'assistant',
  mistral.completionModel('mistral-large-latest'),
).build()

const result = await agent.prompt('Extract the launch risks from this report.').send()
console.log(result.output)

Capabilities

Capability	Factory	Default
Streaming completion	`completionModel()`	`mistral-large-latest`
Dense embeddings	`embeddingModel()`	`mistral-embed`
OCR	`ocrModel()`	`mistral-ocr-latest`
Model inventory	`listModels()`	Provider model list

Completion supports text, streaming, tools, tool choice, and structured output. Chat image input, chat document input, transcription, audio generation, and image generation are not currently model factories in this package.

Common patterns

Extract a remote document with OCR

const ocr = mistral.ocrModel()

const result = await ocr.ocr({
  source: {
    type: 'document_url',
    url: 'https://example.com/invoice.pdf',
    documentName: 'invoice.pdf',
  },
  tableFormat: 'markdown',
})

console.log(result.markdown)
console.log(result.pages)

Byte sources are uploaded before OCR and expose upload metadata in uploadedFile. URL and existing file-ID sources avoid that upload step.

Batch embeddings

const embeddings = mistral.embeddingModel('mistral-embed', {
  maxBatchSize: 32,
})

const vectors = await embeddings.embedTexts(['first document', 'second document'])

Compatibility

@anvia/mistral is ESM and uses the official @mistralai/mistralai client. Supply serverURL for a custom service or inject a preconfigured Mistral client. Exported request/response mapping helpers are available for custom adapters, but they still expect the Anvia and Mistral data shapes documented in the API reference.

@anvia/mistral ​

Install ​

Create a completion model ​

Capabilities ​

Common patterns ​

Extract a remote document with OCR ​

Batch embeddings ​

Compatibility ​

Continue ​