Qwen3 Embedding
qwen/qwen3-embeddingMultilingual embedding model priced for indexing everything, the budget default for RAG at scale.
32K ctx · Price on request
Category
Embeddings
Text-to-vector for search and RAG. The invisible workhorse behind semantic search, RAG and recommendations. Reach for an model whenever retrieval comes before generation. Pricing is (there is no generated output to bill), and changing model later means re-indexing everything you have stored.
3 models in Embeddings from 3 providers
R/1M tokens These models bill on input alone. Nothing is generated in return, so the output price on each card reads “—”.
Top models
Leading embedding models
Qwen3 Embedding by Alibaba Qwen Alibaba QwenOn request Gemini Embedding by Google GoogleOn request Gemini Embedding
google/gemini-embeddingHigh-recall embeddings backed by Gemini's multilingual foundation.
8K ctx · Price on request
Embed v4 by Cohere CohereOn request Embed v4
cohere/embed-v4Multimodal enterprise embeddings: text and images in one vector space, 128K inputs.
128K ctx · Price on request
Browse all embedding models →
3 models in the directory
How to choose
Choosing a model in Embeddings
- 01
Input-only pricing
Embeddings bill on input alone: price per 1M tokens, so comparisons are straightforward here.
- 02
Dimensions drive storage
Bigger vectors cost more to store and search. Smaller dimensions with good recall often win in production.
- 03
Keep it stable
Changing embedding models means re-indexing everything. Choose one you can live with.