Qwen3-Embedding-0.6B is now available in Databricks Model Serving, delivering state-of-the-art multilingual embeddings for vector search and AI agents.

databricks

Databricks has launched Qwen3-Embedding-0.6B in public preview on its platform. Built on the Qwen3 foundation, this 0.6B-parameter multilingual embedding model supports up to 32k token context length, covers 100+ languages, and outperforms flagship models from OpenAI and Cohere on MTEB benchmarks while rivaling much larger 7B+ models. It supports Matryoshka Representation Learning (MRL), allowing embeddings to be truncated from 32 to 1024 dimensions at request time for cost/recall tradeoffs. The model runs on secure serverless GPUs within Databricks and integrates with Agent Bricks and Vector Search for building RAG pipelines and AI agents directly on enterprise data. It is available across all clouds via Pay-Per-Token, AI Functions, and Provisioned Throughput.

SOTA Embedding Model for Agentic Workflows Now in Public Preview