Back to Arena

Jina AI Embeddings

by Jina AI GmbH

System Card

OrganizationJina AI GmbH
Released2020-06
Architectureexternal-memory-network / long-context multimodal neural search
DetailsJina AI provides embedding APIs with 8K token context windows (jina-embeddings-v3), enabling document-level dense retrieval without chunking. jina-embeddings-v4 (2025) is a 3.8B-parameter unified text+image model trained with contrastive learning. Also offers Jina Reader (web scraping to clean text) and Jina Reranker. Over 400,000 registered users.
Parameters—
Domainrag-retrievallong-context
Open SourcePartial
WebsiteVisit
long-contextmultimodalrerankingreader-apicontrastive-learning

Capability Profile

Benchmark Scores

6 of 14 benchmarks
Long-Context Retrieval
4/5
RULER
76.189p
NIAH
no data
LooGLE
78.973p
LongBench
603p
∞Bench
83.181p
Multi-Turn Recall
0/2
LoCoMo
no data
MemoryBank
no data
Cross-Session Memory
0/1
LongMemEval
no data
Multi-Hop QA
2/3
BABILong
83.699p
MultiHop-RAG
no data
HotpotQA
66.134p
Agent Task Memory
0/1
AgentBench-Mem
no data
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Jina AI Embeddings vendor documentation; evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)Jina AI Embeddings vendor documentation; evaluated on RULER: What's the Real Context Size of Your Long-Context Language Models (NVIDIA, 2404)Jina AI Embeddings vendor documentation; evaluated on BABILong: Testing the Limits of LLMs with Long-Context Reasoning-in-a-Haystack (AIRI, 2406)Jina AI Embeddings vendor documentation; evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Jina AI Embeddings vendor documentation; evaluated on InfiniteBench: Extending Long Context Evaluation Beyond 100K Tokens (Tsinghua / OpenBMB, 2402)Jina AI Embeddings vendor documentation; evaluated on LooGLE: Can Long-Context Language Models Understand Long Contexts? (Peking University, 2311)