NV-Embed-v2

Name: NV-Embed-v2
Author: nvidia

by nvidia

Top-ranked 7B text embedding model on MTEB English benchmark

1.5Mdl/month

7Bparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

nvidia/NV-Embed-v2

Feature URI

mixpeek://text_extractor@v1/nvidia_nv_embed_v2

Overview

NV-Embed-v2 is NVIDIA's 7B-parameter text embedding model that held the #1 position on the MTEB English benchmark with a score of 72.31. It uses a latent attention layer to remove the mean token pooling bottleneck and applies a two-stage contrastive training recipe: first on retrieval datasets, then on a blend of retrieval plus non-retrieval tasks (classification, clustering, STS).

On Mixpeek, NV-Embed-v2 is the highest-accuracy text embedder available for English-dominant workloads. Its 7B parameter count delivers superior quality for knowledge bases, legal corpora, and technical documentation where recall matters more than latency.

Architecture

Mistral-7B decoder backbone with a learned latent attention pooling layer replacing mean pooling. 7B parameters. Two-stage instruction-tuned contrastive training with causal attention masks removed during embedding.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so NV-Embed-v2 runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns_your_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection_id: "col_your_collection",
      documents: [
        {
          document_id: "asset-00412",
          // The vector name has to match a vector index on the collection.
          vectors: { "text-embedding": yourVector },
          payload: { source_key: "archive/2026/asset-00412" },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// text_extractor@v1 runs intfloat/multilingual-e5-large-instruct
// (1024-d) over a bucket, with no inference of your own.