Pinecone is the trusted AI knowledge company. The homepage meta names Database, Nexus, and Marketplace for 10,000+ customers and 1M developers. You create your first index for free, then pay as you go. Builder is a flat monthly fee. Standard and Enterprise add a monthly minimum. The homepage says to install the plugin in your agent, or call the API directly, then search, upsert, and rerank from the same index.
Official website · www.pinecone.io
Pinecone fits a team that wants a managed index and will read the four plan columns before leaving the free caps. The homepage says Pinecone is a fully managed vector database built for AI, that writes are instantly searchable, and that indexing is automatic. It is a weak fit if you need a chat-model router or a model-file host. The Nexus token, speed, and accuracy lines are the homepage's benchmark sentences.
The homepage says Pinecone is a fully managed vector database built for AI. Writes are instantly searchable, indexing is automatic, and queries stay fast at any scale. A write is acknowledged in under 100ms and searchable within seconds. A query searches all data in parallel. The docs overview says Pinecone is the vector database for AI agents and applications, built for semantic search, knowledge retrieval, and long-term memory at scale. The homepage line is to search, upsert, and rerank from the same index.
Official documentationCreate your first index for free, then pay as you go. Starter is Start for Free, for small applications: database on-demand, inference, assistant, dense, sparse, and full-text indexes, console metrics, and Discord support. Builder is $20/month flat for solo developers and small teams. It says everything in Starter, higher limits, choose your cloud and region, multiple projects and users, and Prometheus and Datadog monitoring. Standard charges a minimum of $50/month, then pay as you go, and a 3 week trial includes $300 credits. It adds dedicated read nodes, object-storage import, backup and restore, RBAC, SSO (SAML 2.0), and a HIPAA add-on. Enterprise charges a minimum of $500/month, then pay as you go, and lists a 99.95% uptime SLA, bring your own cloud, private endpoints, customer managed encryption keys, audit logs, service accounts, SAML roles, SCIM, HIPAA compliance, and pro support.
Official documentationColumns are Starter, Builder, Standard, Enterprise. Cloud reads AWS, then AWS, Azure, and GCP. Region reads us-east-1, then all available regions. Indexes read up to 5, 10 per project, 20 per project, and 200 per project. Namespaces read 100, 1,000, 100,000, and 1,000,000. Reranking reads bge-reranker-v2-m3 only for the first two columns and all available models after that. Storage reads up to 2 GB, up to 10 GB, then unlimited at $0.33/GB/mo on Standard and Enterprise. Writes read up to 2M/mo, up to 5M/mo, then $4-$4.50 per million on Standard and $6-$6.75 per million on Enterprise. Reads read up to 1M/mo, up to 2M/mo, then $16-$18 per million on Standard and $24-$27 per million on Enterprise. The page says those rates vary by cloud and region.
Official documentationNexus compiles enterprise data into governed knowledge once, then serves it through a single query. Across patent search, M&A due diligence, and revenue intelligence, it says Nexus used 85–97% fewer tokens per query than agentic RAG, and that one query was 30x faster in time-to-completion. An LLM-judge line says 14–94% more accurate, largest on 217 call transcripts. Governed access lists encryption at rest and in transit, SSO, RBAC, CMEK, and private networking. The compliant line says SOC 2 Type II, HIPAA, GDPR, and ISO 27001 certified. Docs say to use Pinecone with Claude Code, Codex, Gemini CLI, Cursor, and other agentic tools. The plugin install shown on the homepage is claude plugin install pinecone, then /pinecone:quickstart.
Official documentationA suggested first project stays on Starter: AWS, us-east-1, up to five indexes, up to 2 GB, up to 1M reads a month, and up to 2M writes a month. The page's semantic-search sketch assumes about 30K documents and then says that shape is about 15k searches a day and about 660 document updates a day. The page says the sketches are illustrative only, not quotes or binding offers, and that they exclude inference, assistant, and the initial import.
Point is leaving us-east-1 or needing more than one project. Builder is the $20/month card that says choose your cloud and region, with storage up to 10 GB. Standard is the card for dedicated read nodes, backup and restore, RBAC, or SSO (SAML 2.0), at a $50/month minimum and $300 in trial credits. Enterprise includes a $500/month minimum, a 99.95% uptime SLA, and bring-your-own-cloud in your VPC without SSH, VPN, or inbound access. The suggestion is which column matches the constraint.
A suggested setup follows the plugin page. Export a Pinecone API key, run claude plugin install pinecone, restart Claude Code, and use /pinecone:quickstart. The homepage says official Anthropic marketplace access is required. Other commands on the plugin page include /pinecone:query and /pinecone:assistant. That page also asks for Node.js with npx on PATH and uv for the bundled Python scripts. The CLI is optional.
Start for free and create your first index, then pay as you go when you scale.
Install the plugin in your agent, or call the API directly. The sample command is claude plugin install pinecone.
Search, upsert, and rerank from the same index. Docs also name hybrid search and metadata filters.
For another region, SSO, or backups, read Builder, Standard, or Enterprise before you rely on that card.
Starter is free. Builder is $20/month flat. Standard is a $50/month minimum and includes $300 trial credits. Enterprise is a $500/month minimum. After a minimum, the card says you pay as you go. Unlimited storage is $0.33/GB/mo. Standard writes are $4-$4.50 per million and reads are $16-$18 per million. Enterprise writes are $6-$6.75 per million and reads are $24-$27 per million. Anything over $50 is billed pay-as-you-go.
Trying an index inside the shown caps.
Free.
Start for free.
Solo developers who need another cloud or region.
$20/month.
Flat fee.
Production apps that need backups, RBAC, or SSO.
$50/month.
Minimum, then pay as you go.
Teams that need the SLA, BYOC, or audit logs.
$500/month.
Minimum, then pay as you go.
Starter daily counts are illustrative, not binding offers, and they exclude inference and assistant usage. Nexus percentages are the homepage's benchmarks. Review pages are not used for prices.
You need a hub for model files, not this index bill.
Hugging Face is a separate listing for hosting and sharing models. Pinecone's public pages are an index and a fee table. Choose the hub when the repo is the work.
You need a route across chat models, not a vector index.
OpenRouter's pages describe a model router and credits for calls. Pinecone's pages describe indexes and read and write units. Pick the router when you are not storing vectors. No OpenRouter price is copied here.
You want a hosted model run rather than an index.
Pinecone's homepage is search, upsert, and rerank on an index. Choose Replicate when that index is not the job.
Builder is $20/month flat. Standard is a $50/month minimum, then pay as you go, and a 3 week trial includes $300 credits. Enterprise is a $500/month minimum, then pay as you go. Starter says Start for Free. A usage line says anything over $50 is billed pay-as-you-go.
SourceThe comparison row lists AWS, us-east-1, up to 5 indexes, 100 namespaces, up to 2 GB, up to 2M writes a month, and up to 1M reads a month. Reranking is bge-reranker-v2-m3 only. Support is Discord. The card shows no dollar minimum.
SourceStandard adds dedicated read nodes, object-storage import, backup and restore, RBAC, SSO (SAML 2.0), and a HIPAA add-on. Enterprise adds a 99.95% uptime SLA, bring your own cloud, private endpoints, customer managed keys, audit logs, service accounts, SAML roles, SCIM, HIPAA compliance, and pro support.
SourceA write is acknowledged in under 100ms and searchable within seconds. Indexing is automatic. A query searches all data in parallel. Docs quickstarts add semantic search, an assistant over proprietary data, and a Nexus context with grounded, cited answers.
SourceSet a Pinecone API key, run claude plugin install pinecone or /plugin install pinecone, restart, and run /pinecone:help. You need marketplace access, Node.js with npx on PATH, and uv. The homepage sample ends at /pinecone:quickstart.
SourceMarketplace is named in the meta line and not described further here. The 85–97% token line, the 30x line, and the 14–94% accuracy line are homepage benchmarks. No model-training rule is stated. The compliant line says SOC 2 Type II, HIPAA, GDPR, and ISO 27001 certified.
SourceOct 6, 2026 · NavHunt editorial
NavHunt did not run Pinecone, did not create an index, and did not send a search. This profile restates the homepage, pricing table, docs overview, and Claude Code plugin page reviewed on 6 October 2026. Benchmark percentages and Starter examples stay the vendor's text. This description is based on Pinecone's official materials reviewed on 2026-10-06, including Pinecone homepage, Pinecone pricing, Pinecone docs overview, Pinecone Claude Code plugin.
Compare the free index cap, the $20 Builder fee, and the $50 and $500 monthly minimums on the public pricing page.