Problem using auto embedding with jinaai/jina-embeddings-v5-text-nano

General Discussions - English
1

Hello,

I get error when trying to create table using auto embedding with model jinaai/jina-embeddings-v5-text-nano.

CREATE TABLE vector_test (
query TEXT,
vector FLOAT_VECTOR KNN_TYPE=‘hnsw’ HNSW_SIMILARITY=‘l2’
MODEL_NAME=‘jinaai/jina-embeddings-v5-text-nano’
FROM=‘query’
)

Error in query (1064): error adding table ‘vector_test’: prealloc: Failed to parse model configuration

Model Qwen/Qwen3-Embedding-0.6B works well.

On this page Jina is mentioned

How to know what model is supported?

Thank you in advance!

2

What was meant in the article here:

  • :rocket: Voyage & Jina models — domain- and language-optimized

was that you who can use Jina models via their API, e.g., like in this test manticoresearch/test/clt-tests/mcl/auto-embeddings-jina-remote.rec at e605e80c5157b7630a6839d1bf03c27f082c1b03 · manticoresoftware/manticoresearch · GitHub?

mysql -h0 -P9306 -e "CREATE TABLE test_jina_remote (title TEXT, content TEXT, description TEXT, embedding FLOAT_VECTOR KNN_TYPE='hnsw' HNSW_SIMILARITY='l2' MODEL_NAME = 'jina/jina-embeddings-v4' FROM = 'title, content' API_KEY='${JINA_API_KEY}') "; echo $?

How to know what model is supported?

See Creating a table > Data types | Manticore Search Manual and Searching > KNN | Manticore Search Manual