Embedding model

codestral-embed

Mistral code embedding model tuned for code search, repository retrieval, and coding-assistant context recall. Token IDs on this page use the shared Mistral Tekken v3 tokenizer vocabulary.

Mistral

Endpoint
Embeddings
Status
Current
Default dimensions
1,536
Max input
8,192 tokens
Tokenizer
mistral_tekken_v3
Tokenizer tokens
131,072 known tokens (130,072 mergeable)
Open model token IDs Open tokenizer reference Mistral docs