Pinecone Inference
pinecone.io · 2 records in Databases and KV, Inference · JSON
Records 2
- caveatPineconeUSD 330/TB-month
Databases and KVfree tieridentity emailno cardlimit request-rateevidence documentedstale on 2026-11-22
- caveatPinecone Inference (embeddings and reranking)
Inferencefree tieridentity emailcard?limit volume-quotaevidence documentedstale on 2026-12-04
Pinecone
Databases and KV- Free allowance
- Starter: 2 GB serverless storage, 5 indexes, 1 project, 1 GB egress per month, us-east-1 only
- Verdict
- caveatusable, with a named limit or an unprobed read path
- Cost
- free tierpaid plans from USD 330/TB-month
- Identity
- emailno card, no phone
- Binding limit
- request-rate1 GB egress per month on Starter; 5 indexes, 1 project per organization, us-east-1 only, and 100 namespaces per index; no connection limit is published for Starter; recorded as a gap
- Evidence
- documented1 source; a first-party page was read and the sentence quoted
- Freshness
- stale on 2026-11-22checked 2026-10-22 (live); stale on 2026-11-22
| Free tier | Starter: 2 GB serverless storage, 5 indexes, 1 project, 1 GB egress per month, us-east-1 only unit GB; metered on stored-bytes |
|---|---|
| Price | USD 0.33/GB-month storage; USD 0.10/GB egress |
| Limits |
|
| Identity | email email signup; no card on Starter |
| Egress | 1 GB per month on Starter. Builder is 10 GB; Standard includes 100 GB then USD 0.10/GB free allowance: 1 GB per month price per tb: USD 0.10/GB on Standard |
| Per-object cap | 2 GB of index storage per organization; 100 namespaces per index |
| Retention | an index persists while the plan is active; no deletion clock is published for Starter |
| Interface | rest, cli a REST API for data and control planes; Python, Node and Java SDKs; the pinecone CLI |
| Tooling | the Pinecone SDKs and CLI; no Postgres extension path |
| Composes with | a free vector index that mostly matters for a workload too large for pgvector on a free Postgres, because pgvector is already available on both Neon and Supabase at no extra cost |
| Reliability | no reliability event recorded |
| Notes | 2 GB is an organization-level cap and one project is the whole allowance, so there is no way to shard around it. 1 GB a month of egress is 33 MB a day, which a retrieval workload serving embeddings exceeds before the 2 GB of storage matters. |
| Evidence note | documented the pricing comparison table was read. The pricing page states them and is now the only source for them. |
| Parsed numbers |
|
Evidence 1
| date | source | what it proved |
|---|---|---|
| 2026-10-22 | www.pinecone.io/pricing/ | Starter: storage up to 2 GB, egress up to 1 GB/month, indexes up to 5, namespaces per index 100; Builder USD 25 with 10 GB storage and 10 GB egress; Standard unlimited at USD 0.33/GB-month |
Unknown values 1
- reliability note: nothing of this kind was observed
- Free allowance
- Starter plan, labelled Free: 5,000,000 embedding tokens a month per model (llama-text-embed-v2, multilingual-e5-large, pinecone-sparse-english-v0, text-embedding-3-small and -large) and 500 rerank requests a month (bge-reranker-v2-m3, pinecone-rerank-v0)
- Verdict
- caveatusable, with a named limit or an unprobed read path
- Cost
- free tier
- Identity
- emailcard unknown, no phone
- Binding limit
- volume-quotaStarter: 5,000,000 embedding tokens per month per model and 500 rerank requests per month per rerank model; reaching either returns HTTP 429
- Evidence
- documented10 sources; a first-party page was read and the sentence quoted
- Freshness
- stale on 2026-12-04checked 2026-11-03 (live); stale on 2026-12-04
| Free tier | Starter plan, labelled Free: 5,000,000 embedding tokens a month per model (llama-text-embed-v2, multilingual-e5-large, pinecone-sparse-english-v0, text-embedding-3-small and -large) and 500 rerank requests a month (bge-reranker-v2-m3, pinecone-rerank-v0) unit tokens; metered on request-count; the pricing page lists the three Pinecone-hosted embedding models and bge-reranker-v2-m3 only; the limits page also lists text-embedding-3 models and pinecone-rerank-v0 for Starter; gemini-embedding-2 and cohere-rerank-4-fast are not available on Starter |
|---|---|
| Price | none stated |
| Limits |
|
| Binding | volume-quota 500 rerank requests a month binds first on the rerank side; 5M tokens a month per model on the embedding side |
| Identity | email a Pinecone account; the pricing page labels Starter 'Free' and its start link goes to the billing settings of the console; unverified: the sign-up form fields and whether a card is asked, because the console page returned no content to the page reader |
| Egress | not published |
| Interface | rest endpoint: https://api.pinecone.io/embed |
| Tooling | Pinecone SDKs and the REST inference endpoints embed and rerank; embeddings can be called without creating an index |
| Composes with | the Starter plan's index allowance (5 indexes, 2 GB) is the same account's vector store, so embed, store and rerank stay under one key |
| Reliability | no reliability event recorded; the Starter allowance is per month and resets |
| Notes | derived for scale: 500 rerank requests a month is about 16 a day, so rerank here suits a prototype or a nightly batch |
| Evidence note | documented the monthly token and request allowances and the per-minute limits are quoted from the vendor's limits page and pricing page; the sign-up form fields are unverified. The two pages disagree on pinecone-rerank-v0 for Starter. |
| Parsed numbers |
|
Evidence 10
| date | source | what it proved |
|---|---|---|
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | Starter allows 5,000,000 embedding tokens a month per model |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | Starter allows 500 rerank requests a month on bge-reranker-v2-m3 |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | Starter allows 500 rerank requests a month on pinecone-rerank-v0, which the pricing page does not list for Starter |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | Starter passage embedding throughput on llama-text-embed-v2 is 250,000 tokens per minute |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | bge-reranker-v2-m3 is limited to 60 requests per minute on Starter |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | inference requests are limited to 2,000 per minute per project on every plan |
| 2026-11-03 | docs.pinecone.io/reference/api/database-limits/rate-limits.m | the monthly limits answer HTTP 429 |
| 2026-11-03 | www.pinecone.io/pricing/ | the pricing page states 5M tokens a month included on Starter for llama-text-embed-v2 |
| 2026-11-03 | www.pinecone.io/pricing/ | the pricing page states 500 rerank requests a month included on Starter |
| 2026-11-03 | www.pinecone.io/pricing/ | Pinecone Inference is part of the Starter plan, which the page labels Free |
Unknown values 3
- egress metered: egress metering was not published
- per object cap stated: the field does not apply to this kind of offer
- retention stated: the field does not apply to this kind of offer