pinecone-rerank-v0
AI model · Reranker / retrieval · Text → Tables / structured data by Pinecone
Enhance search precision with Pinecone's native reranking model.
A specialized text reranking model by Pinecone designed to enhance search relevance with a 512-token context window. It is available via a standalone API endpoint for developers building retrieval-augmented generation (RAG) systems.
Model specs
The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.
- Languages
- English
Category: AI Infrastructure
- Modalities
- Text → Reranked Text
- Context window
- 512 tokens
- Max input
- 512 tokens
- Access
- API (/rerank endpoint)
- Languages
- English
- Type
- Reranking
- 512 token context window for text reranking
- Available via standalone /rerank API endpoint
- Processed in US-based infrastructure
- Accessible on Pinecone Starter plan and higher
Overview
pinecone-rerank-v0 is a specialized reranking model developed by Pinecone to improve the precision of search results in AI applications. It serves as a second-stage retrieval component that re-evaluates an initial set of documents against a query to surface the most relevant information. The model supports a context window of 512 tokens and is processed within US-based infrastructure. While currently marked as deprecated in the Pinecone model gallery, it remains accessible via a standalone /rerank API endpoint. This allows developers to integrate reranking into their workflows with or without a Pinecone index, provided they are on at least the Starter plan.
Model Specifications
- 512 Token Context Window: The model supports an input limit of 512 tokens for reranking operations.
- Text Input Modality: It processes text inputs to re-order search results based on relevance.
- US Processing Location: The model is hosted and processed in the United States.
Access and Integration
- Standalone Reranking API: The model can be accessed via the standalone /rerank endpoint to rerank results without an index.
- Legacy Records API Support: It is available for use with the legacy Records API for integrated indexes.
- Starter Plan Access: The model is accessible to users on the Pinecone Starter plan and higher.
Use cases
- Developers use the model to re-order initial search results to ensure the most relevant documents appear at the top of the list.
- AI teams integrate the /rerank endpoint into RAG pipelines to improve the quality of context provided to large language models.
- Engineers use the model to refine keyword-based search results with deeper semantic understanding.
Specifications
- Modalities
- Text → Reranked Text
- Context window
- 512 tokens
- Max input
- 512 tokens
- Access
- API (/rerank endpoint)
- Languages
- English
- Type
- Reranking
- Company
- Pinecone
- Input Modality
- Text
- Max Tokens
- 512 tokens
- Processing Location
- US
- Status
- Deprecated
How it works
- Submit Query and Documents: The user provides a query and a list of documents to the standalone /rerank API endpoint.
- Semantic Processing: The model processes the text inputs in US-based infrastructure to determine semantic relevance.
- Receive Reranked Results: The API returns the documents re-ordered by their relevance scores to the original query.
Pricing
- Starter Plan and higher
- Included in plan usage limits
FAQ
Is pinecone-rerank-v0 still supported?
Yes, pinecone-rerank-v0 is currently listed as deprecated in the Pinecone model gallery.
What is the context window for this model?
The model supports a maximum of 512 tokens per input.
Which Pinecone plan is required to use this model?
It is available starting from the Pinecone Starter plan.
Where is the data for this model processed?
The model is processed in the United States.
Added by AntSpace from public sources (docs.pinecone.io).