Jina CLIP v1
Jina CLIP v1 aligns English text and images while also supporting text-to-text retrieval. Its training addresses a weakness of conventional CLIP models, whose image alignment does not automatically make them strong text retrievers.
How the model works
The model combines text and vision encoders in a shared representation. Its card compares both cross-modal and text-only retrieval, rather than relying on image classification alone. Jina later expanded the design with multilingual text, larger image inputs and adjustable vector width in CLIP v2.Jina CLIP v1 model details
Use and access
The v1 weights use Apache 2.0 and include Transformers, Sentence Transformers and JavaScript examples. The model card warns that text-text similarity tends to be larger than text-image similarity. An application that mixes these scores should therefore evaluate its combination rule instead of assuming they have the same scale.Jina CLIP v1 model detailsJina CLIP v1 technical overview
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | Downloadable |
| Model identifiers | jina-clip-v1jinaai/jina-clip-v1 |
| Weight license | Apache 2.0Jina CLIP v1 model details |
Related models
- Jina CLIP v2
Jina AI · Jina CLIP
A multilingual text-and-image encoder with adjustable vector dimensions.
Launch date API · Downloadable - Jina Embeddings v2 Base English
Jina AI · Jina Embeddings v2
An English long-document encoder with permissively licensed weights.
Launch date Downloadable