inputs
Required
Type: Text
Supplies the text whose meaning will be represented as numeric vectors. Use a JSON list to embed multiple items in one request; each input produces its own vector for search, clustering, or similarity comparison.
ChinaEmbedding100+ languages32K context
Explore Qwen3-Embedding-8B
Create multilingual text embeddings for semantic search, retrieval, clustering, classification, and code search with configurable Matryoshka dimensions. Qwen3-Embedding-8B is developed by Alibaba. Adjust the available settings to shape the result for your task.
Model
Qwen/Qwen3-Embedding-8B
Documentation
Qwen3-Embedding-8B field reference
Use this reference to understand how every field in the playground changes the request sent to Qwen/Qwen3-Embedding-8B.
truncation
Optional
Type: Boolean
Default: On
Allows the tokenizer to discard tokens when an input is longer than the configured maximum or model context. Turning it off surfaces an error instead of silently losing content, which is safer when every part of the input matters.
tokenize_kwargs
Optional
Type: JSON object
Default: None
Provides advanced keyword arguments to the tokenizer before embeddings are computed. Use it to control padding, truncation, or maximum length; truncating too aggressively can remove text that is important to semantic similarity.
return_tensors
Optional
Type: Boolean
Default: Off
Returns numeric generated token IDs as tensors rather than ordinary readable text. This is useful for downstream model processing or custom decoding, but most applications should leave it off.