https://pinecone.io

The vector database to build knowledgeable AI | PineconeSearch through billions of items for similar matches to any object, in milliseconds. It’s the next generation of search, an API call away.The vector database to build knowledgeable AI | PineconeAnnouncementPinecone Assistant is now in public preview — simplify, enhance, and evaluate RAG developmentLearn moreProductPricingResources Company DocsLoginSign upBuild knowledgeable AIPinecone serverless lets you deliver remarkable GenAI applications faster.Get StartedContact SalesBuild knowledgeable AI applications in minutesQuickstartStart building with a serverless vector database todayCreate an account and your first index in 30 seconds, then upload a few vector embeddings from any model… or a few billion.Get StartedGenerate your API key and implement in seconds.Generate KeyDocs # Install these packages: # !pip install pinecone # !pip install pinecone_datasets from pinecone import Pinecone, ServerlessSpec from pinecone_datasets import load_dataset # Initialize Pinecone pc = Pinecone(api_key="YOUR_API_KEY") # Create the index if it doesn't already exist pc.create_index(name="hello-pinecone", dimension=1536, spec=ServerlessSpec(cloud='aws', region='us-east-1')) # Connect to the index index = pc.Index(name='hello-pinecone') # Load the dataset dataset = load_dataset("langchain-python-docs-text-embedding-ada-002") # Define a function to upsert batches def upsert_batch(batch): index.upsert(vectors=batch) print("Your index has been created! Uploading a few items from the dataset...") # Upsert the first 1000 entries from the dataset for i, batch in enumerate(dataset.iter_documents(batch_size=50)): index.upsert(vectors=batch) if (i + 1) * 50 >= 1000: break # Function to get the first entry def get_first_entry(dataset): for batch in dataset.iter_documents(batch_size=1): return batch[0]['values'].tolist() # Query the index print(index.query( vector=get_first_entry(dataset), top_k=1, include_values=True, include_metadata=True )) Pinecone is the vector database that helps power AI for the world’s best companiesStart and scale seamlesslyStartSearchScale Create an account and your first index in 30 seconds, then upload a few vector embeddings from any model… or a few billion. Perform low-latency vector search to retrieve relevant data for search, RAG, recommendation, detection, and other applications. Pinecone is serverless so you never have to worry about managing or scaling the database. Quickstart Guide >PythonNodecURLJavaDocs from pinecone import Pinecone, ServerlessSpec # Create a serverless index # "dimension" needs to match the dimensions of the vectors you upsert pc = Pinecone(api_key="YOUR_API_KEY") pc.create_index(name="products", dimension=1536, spec=ServerlessSpec(cloud='aws', region='us-east-1') ) # Target the index index = pc.Index("products") # Mock vector and metadata objects (you would bring your own) vector = [0.010, 2.34,...] # len(vector) = 1536 metadata = {"id": 3056, "description": "Networked neural adapter"} # Upsert your vector(s) index.upsert( vectors=[ {"id": "some_id", "values": vector, "metadata": metadata} ] ) SearchScaleSearch and scale seamlesslyPerform low-latency vector search to retrieve relevant data for search, RAG, recommendation, detection, and other applications.Quickstart GuideDocs # Mock vectorized search query (vectorize with LLM of choice) query = [0.13, 0.45, 1.34, ...] # len(query) = 1536, same as the indexed vectors # Send query with (optional) filter to index and get back 1 result (top_k=1) index.query( vector=query, filter={"description": {"$eq": "Networked neural adapter"}}, top_k=1 ) More relevant results make better applicationsFilter bymetadataCombine vector search with familiar metadata filters to get just the results you want.FindcontextFast and accurate vector search over all your data.Update inreal timeAs your data changes, the Pinecone index is updated in realtime to provide the freshest results.Make (the right)keywords matterCombine vector search with keyword boosting for the best of both worlds (hybrid search).30k+organizations96%recall*51msquery latency (p95)**Performance with MSMarco V2 dataset of 138M embeddings (1536 dimensions)Part of the developer-favorite AI stackUse Pinecone with your favorite cloud provider, data sources, models, frameworks, and more.Explore integrations SearchGenerateFine TuneData sourceEmbedding modelPinecone Vector DatabaseSearch applicationJoin the movementJoin a growing community of 400,000+ ambitious developers building the next generation of applications with Pinecone.EventsLearn and connect with your peers, in person and online.Attend an event >DocsTake advantage of our developer-friendly docs to get going in minutes.Quickstart >ForumShare your questions, and answers in the support forum.Ask the community >Billionsof records in PineconeNotion is leading the AI productivity revolution. Our launch of a first-to-market AI feature was made possible by Pinecone serverless. Their technology enables our Q&A AI to deliver instant answers to millions of users, sourced from billions of documents. Best of all, our move to their latest architecture has cut our costs by 60%, advancing our mission to make software toolmaking ubiquitous.Akshay KothariCo-Founder and COO, NotionSecure and Enterprise-readyMeet security and operational requirements to bring AI products to market faster.SecureControl your data and know it’s safe. Pinecone is SOC 2 and HIPAA certified.ReliablePowering mission-critical applications of all sizes, with support SLAs and observability.Cloud-nativeFully managed in the cloud of your choice. Also available via marketplaces: AWS, Azure, GCP.Explore security Start building knowledgeable AI nowCreate your first index for free, then upgrade and pay as you go when you're ready to scale, or talk to sales.Get StartedProductOverviewDocumentationIntegrationsTrust and SecuritySolutionsCustomersRAGSemantic SearchMulti-Modal SearchCandidate GenerationClassificationResourcesLearning CenterCommunityPinecone BlogSupport CenterSystem StatusWhat is a Vector Database?What is Retrieval Augmented Generation (RAG)?Multimodal Search WhitepaperClassification WhitepaperCompanyAboutPartnersCareersNewsroomContactLegalCustomer TermsWebsite TermsPrivacyCookiesCookie Preferences© Pinecone Systems, Inc. | San Francisco, CAPinecone is a registered trademark of Pinecone Systems, Inc.enen1727289636https://pinecone.io

Fa'atonu lau saite?

O au mea na e fai?