Jina Reranker v1 Turbo English
The turbo v1 reranker is an English relevance model with 37.8 million parameters and 6 transformer layers. Jina distilled it from the larger v1 base teacher to reduce inference cost.
How the model works
Turbo uses six JinaBERT layers with a 384-wide hidden state, retaining two more layers than Tiny. Symmetric bidirectional ALiBi supports 8,192-token input. Knowledge distillation transfers relevance behavior from the 137-million-parameter base model into this smaller checkpoint.Jina Reranker v1 Turbo English model details
Use and access
The downloadable weights use Apache 2.0. Jina documents Python and JavaScript inference as well as its managed reranking service. This is a historical English release with accessible weights; later multilingual models do not make it a different model or remove its license. It returns relevance scores, not generated answers.Jina Reranker v1 Turbo English model detailsJina Reranker v1 Turbo English technical overview
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | jina-reranker-v1-turbo-enjinaai/jina-reranker-v1-turbo-en |
| Context window | 8,192 tokens Maximum input sequence in the model card; serving limits can be lower. Jina Reranker v1 Turbo English model details |
| Parameters | 37.8 millionJina Reranker v1 Turbo English model details |
| Weight license | Apache 2.0Jina Reranker v1 Turbo English model details |
| Base model | JinaBERT; distilled from v1 Base EnglishJina Reranker v1 Turbo English model details |
Related models
- Jina Reranker v1 Tiny English
Jina AI · Jina Reranker
A four-layer English ranker distilled for fast relevance scoring.
Launch date Downloadable - Jina Reranker v1 Base English
Jina AI · Jina Reranker
The original English query-document relevance model in Jina's reranking family.
Launch date API