Jina
Jina AI was founded in 2020 by Han Xiao (CEO) and Nan Wang (CTO) as an open-source neural search startup based in Berlin.
Profile
Jina AI provides open-source neural search models that convert text, images, and other unstructured data into vector embeddings for retrieval and reranking, enabling multilingual and multimodal search in AI applications.
Jina AI was founded in 2020 by Han Xiao (CEO) and Nan Wang (CTO) as an open-source neural search startup based in Berlin. The company built a portfolio of open-source models for multimodal and multilingual embeddings, rerankers, and small language models, with its flagship 'serve' repository (formerly Jina) accumulating 21,900 GitHub stars and 8,644 commits as of March 2025. Jina AI raised a $30 million Series A in November 2021 led by Canaan Partners, with participation from Mango Capital, GGV Capital, SAP.iO, and Yunqi Capital.
In October 2025, Elastic (NYSE: ESTC) completed its acquisition of Jina AI for an undisclosed sum, integrating the team and technology into its Search AI Platform to strengthen vector search, retrieval-augmented generation (RAG), and context engineering capabilities. The acquisition added dense vector, multilingual, and multimodal embedding models that process both text and images, complementing Elastic's ELSER model, along with advanced rerankers and small language models for HTML-to-Markdown conversion. Jina AI's customer logos on its website include InVideo AI, Ampeco, ByteDance, and Answear.
The company's financials post-acquisition are not publicly disclosed as a standalone entity; Elastic's most recent quarterly revenue (Q1 FY2027, ended April 2026) was $81.6 billion, up 85% year-over-year, per the Wall Street Journal. Jina AI's models remain open-source and available via GitHub, but the company no longer operates independently.
Who buys this
- Enterprises building search-powered AI applications (e.g., RAG, agentic AI)
- Developers integrating vector search into existing platforms
- E-commerce and retail companies needing multilingual product search
- Media and content platforms requiring multimodal search (text + images)
- AI research teams seeking open-source embedding and reranker models
Publicly disclosed clients
- InVideo AI
- Ampeco
- ByteDance
- Answear
Strengths and what to watch
Strengths
- Acquired by Elastic in October 2025, providing a large distribution channel and enterprise customer base for its models.
- Open-source portfolio with strong community traction (21,900 GitHub stars on serve repo) and broad multilingual/multimodal capabilities.
- Technology directly addresses the growing demand for context engineering in agentic AI and RAG pipelines, a market Elastic is prioritizing.
Watch for
- Post-acquisition, Jina AI's models may become more tightly integrated into Elastic's proprietary platform, potentially reducing their open-source independence.
- The acquisition price was not disclosed, making it difficult to assess the financial return for investors or the valuation at exit.
- Competition in the embedding and reranker model space is intense, with alternatives from Cohere, OpenAI, and open-source models like BGE and GTE.
Recent moves
Key Information
- Industry
- MLOps & AI Infra
- Founded
- 2020
- Headquarters
- Berlin
Frequently Asked Questions
What is Jina AI and what does it do?
Jina AI is an open-source neural search startup founded in 2020 that provides models to convert text, images, and unstructured data into vector embeddings for retrieval and reranking, enabling multilingual and multimodal search in AI applications.
Did Elastic acquire Jina AI?
Yes, Elastic completed its acquisition of Jina AI in October 2025 for an undisclosed sum. The team and technology were integrated into Elastic's Search AI Platform to strengthen vector search, retrieval-augmented generation, and context engineering capabilities.
What are Jina AI's open-source models used for?
Jina AI's open-source models are used for generating dense vector embeddings from text and images, reranking search results, and converting HTML to Markdown. They support multilingual and multimodal search, making them valuable for RAG pipelines and agentic AI applications.
Who are some notable customers of Jina AI?
Jina AI's website lists customers including InVideo AI, Ampeco, ByteDance, and Answear. These companies use Jina AI's models for search-powered AI applications, such as multilingual product search and multimodal content retrieval.
How much funding did Jina AI raise before being acquired?
Jina AI raised a $30 million Series A in November 2021 led by Canaan Partners, with participation from Mango Capital, GGV Capital, SAP.iO, and Yunqi Capital. The acquisition price by Elastic was not publicly disclosed.
What is the GitHub community traction for Jina AI's serve repository?
Jina AI's flagship 'serve' repository, formerly known as Jina, has accumulated 21,900 GitHub stars and 8,644 commits as of March 2025. This reflects strong community engagement with its open-source neural search framework.
Sources
- www.stocktitan.net — Elastic's acquisition of Jina AI completed in October 2025, including details on technology integration and strategic rationale.
- github.com — Jina AI's open-source repository with 21,900 stars, 8,644 commits, and latest version 3.34.0 as of March 2025.
- jina.ai — Customer logos including InVideo AI, Ampeco, ByteDance, and Answear.
- techcrunch.com — Series A funding of $30 million in November 2021 led by Canaan Partners, with details on investors and founding team.
- www.wsj.com — Elastic's Q1 FY2027 revenue of $81.6 billion, up 85% year-over-year, providing context on the acquiring company's financial performance.