nomic-embed-text-v2-moe

Name: nomic-embed-text-v2-moe
Author: nomic-ai

by nomic-ai

First Mixture-of-Experts text embedding model with 100-language multilingual support

1.4Mdl/month

475M total / 305M activeparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

nomic-ai/nomic-embed-text-v2-moe

Feature URI

mixpeek://text_extractor@v1/nomic_embed_v2_moe_v1

Overview

Nomic Embed Text v2 MoE is the first general-purpose Mixture-of-Experts text embedding model, using 8 experts with top-2 routing to deliver 475M total parameters but only 305M active at inference. Trained on 1.6B high-quality pairs with consistency filtering, it achieves state-of-the-art performance on both BEIR and MIRACL benchmarks while remaining competitive with models twice its size.

On Mixpeek, nomic-embed-text-v2-moe provides efficient multilingual text embeddings for search pipelines that span ~100 languages, with Matryoshka dimension support (768 down to 256) for flexible storage and retrieval tradeoffs.

Architecture

Mixture-of-Experts transformer encoder with 8 experts, top-2 routing. 475M total parameters, 305M active during inference. Trained with weakly-supervised contrastive pretraining followed by supervised fine-tuning. Matryoshka representation learning for flexible output dimensions.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so nomic-embed-text-v2-moe 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.