Qwen Text Embedding V4
Awaiting activationSearch multilingual and programming-language content with flexible Qwen3 vectors.
What it does
Alibaba Cloud’s Text Embedding V4 belongs to the Qwen3-Embedding series. The official documentation lists more than 100 languages and programming languages, making it a useful candidate for business information and technical-document retrieval.
It supports eight output sizes, from 64 to 2,048 dimensions, with 1,024 as the default. The provider permits 8,192 tokens per input; this service uses an 8,000-byte cap and batches of up to ten texts to respect the published API contract.
Where it helps
Chinese business search
Match questions with product, policy and service information.
Technical knowledge
Retrieve relevant developer documentation and code-related passages.
Global support
Search help content across the provider’s supported languages.
When to choose it
Consider it for multilingual or technical collections when choosing a compact vector size is valuable. Compare it with newer Qwen 3.7 and code-specialized Voyage models.
Awaiting activation and account-rate confirmation. Changing dimension or model normally requires rebuilding document vectors.
Official sources
Paraphrased from the provider’s documentation and original model information. Selection advice is our editorial guidance.
Alibaba Cloud · Qwen3 Text Embedding V4 ↗Reviewed October 4, 2026