MODELS
nomic-embed-text-v2-moe
by Nomic AI
Overview
Open-source multilingual Mixture-of-Experts text embedding model for retrieval and RAG.
Details
nomic-embed-text-v2-moe is a Nomic AI multilingual Mixture-of-Experts text embedding model. The supplied official sources describe it as open-source and intended for retrieval and RAG, with official weights, code, paper, and a Hugging Face model card. Nomic’s API reference also documents a text-embedding endpoint, POST /v1/embedding/text, with task_type values including search_document, search_query, classification, and clustering.
When to Use
Use for evaluating multilingual text embeddings for retrieval or RAG workflows. Use when you want an open-source embedding model with official weights and code available from Nomic AI. Use Nomic’s hosted embedding API when your workflow needs documented task types such as search_document search_query classification or clustering.
Getting Started
- Review the Hugging Face model card for official weights and model usage details.
- Read Nomic’s launch/blog documentation for model context and intended retrieval/RAG use cases.
- Inspect the nomic-ai/contrastors GitHub repository for the official code source.
- If using hosted inference
- review Nomic’s POST /v1/embedding/text API reference and supported task_type values.
- Check Nomic’s pricing page before relying on the hosted Developer API in production.
Key Features
- •Multilingual text embedding model.
- •Mixture-of-Experts embedding architecture
- •according to Nomic’s product page/source metadata.
- •Positioned for retrieval and RAG use cases.
- •Official Hugging Face model card with weights.
- •Official code repository linked from Nomic AI.
- •Hosted text-embedding API documentation with task types for search
- •classification
- •and clustering.
Capabilities
- •text-embedding
- •multilingual-embeddings
- •semantic-search
- •retrieval
- •rag
- •classification
- •clustering
Last updated Jun 22, 2026