Embedding model

gte-Qwen2-7B-instruct

Higher-capacity Qwen2-based multilingual GTE embedding model for retrieval-heavy and ranking-adjacent semantic workflows. Token IDs on this page use the shared Qwen2 tokenizer vocabulary.

Alibaba

Endpoint

Embeddings

Status

Current

Default dimensions

3,584

Max input

32,000 tokens

Tokenizer

qwen2

Tokenizer tokens

151,646 known tokens (151,643 mergeable)

Open model token IDs Open tokenizer reference Model card