ModelCards

Mistral Embed

Embedding modelCard summarized

Our embedding model has been designed with retrieval capabilities in mind

Source: "La plateforme" launch post and Text Embeddings docs (no model card published)

Use it via API

$0.10 / 1M tok
Code exampleCompare

Where to call it

  • Mistral APImistral-embed

Supports

  • Batch API
Full feature support

Good for

  • Retrieval for RAG
  • Semantic search
  • Clustering and classification

Watch out for

Overview

Mistral AI · 01Embed, front

Mistral Embed

Our embedding model has been designed with retrieval capabilities in mind

Typical price
$0.10 / 1M tok
Text input
Dimensions
1024
Max input
8K tokens
Modes
text embeddings
Released
11 Dec 2023

Mistral Embed is Mistral AI's general-purpose text embedding model, launched with La Plateforme in December 2023. It returns 1024-dimensional vectors and was designed for retrieval; Mistral reported a retrieval score of 55.26 on MTEB at launch. launch post; text embeddings docs

mistral-embed-2312
Typical price
$0.10 / 1M tok
Dimensions
1024
Max input
8K tokens
Weights
Closed
Released
11 Dec 2023
  • text embeddings