Weaviate has launched Weaviate Embeddings, a Software-as-a-Service (SaaS) designed to enhance flexibility in AI development through data vectorization. The service offers a combination of open-source models and managed service convenience with pay-as-you-go pricing.

Embedding services are vital for AI applications, converting data into vector embeddings stored in a vector database. However, they can hinder development due to restrictive rate limits and the necessity for external API calls. Weaviate aims to eliminate these bottlenecks.

The new service allows users to access both open-source and proprietary models hosted in Weaviate Cloud, eliminating the need for external providers. Users maintain full control over their embeddings and can easily switch between models as needed.

Weaviate Embeddings operates on GPU infrastructure, positioning machine learning models near data to reduce latency. The platform imposes no rate limits in production, enabling rapid operations with straightforward pricing that cuts costs for model inference.

CEO Bob van Luijt stated, “Our goal is to equip developers with the tools and operational support to bring their models closer to their data. Weaviate Embeddings makes it simpler for developers to build and manage AI-native applications.” The open-source database supports customization, further providing developers with flexibility.

Initially available in preview within Weaviate Cloud, the service includes Snowflake’s Arctic-Embed, an efficient open-source text embedding model known for its retrieval quality, with future enhancements planned for early 2025.

This latest offering aligns with Weaviate’s mission to assist AI developers in transitioning from prototyping to production environments. Earlier this year, the company launched a workbench of tools for various AI use cases and introduced adaptable storage options to help manage costs for AI-native applications.