Skip to content

Pinecone Compatibility Matrix

Pinecone Compatibility Matrix

Comprehensive compatibility reference for HeliosDB’s Pinecone vector protocol implementation.

Overall Compatibility: 100%

CategoryCoverageResult
REST API100%Complete
Upsert Operations100%Complete
Query Operations100%Complete
Filtering100%Complete
Namespaces100%Complete
Metadata100%Complete

API Endpoints

Index Operations

EndpointMethodResultNotes
/describe_indexGETSupportedIndex statistics
/create_indexPOSTSupportedCreate new index
/delete_indexDELETESupportedDelete index
/list_indexesGETSupportedList all indexes
/configure_indexPATCHSupportedUpdate configuration

Vector Operations

EndpointMethodResultNotes
/vectors/upsertPOSTSupportedInsert/update vectors
/queryPOSTSupportedSimilarity search
/vectors/fetchGETSupportedFetch by ID
/vectors/deletePOSTSupportedDelete vectors
/vectors/updatePOSTSupportedUpdate metadata
/vectors/listGETSupportedList vector IDs

Statistics

EndpointMethodResultNotes
/describe_index_statsGETSupportedIndex statistics

Upsert Operations

FeatureResultNotes
Single vectorSupported
Batch vectorsSupportedUp to 100/request
With metadataSupported
With sparse vectorsSupportedHybrid search
Namespace supportSupported

Query Operations

Query Parameters

ParameterResultNotes
vectorSupportedQuery vector (required)
top_kSupportedNumber of results
filterSupportedMetadata filtering
include_valuesSupportedReturn vectors
include_metadataSupportedReturn metadata
namespaceSupportedQuery namespace

Similarity Metrics

MetricResultNotes
CosineSupportedDefault
EuclideanSupportedL2 distance
Dot ProductSupportedFor normalized vectors

Filtering

Filter Operators

OperatorResultExample
$eqSupportedEqual
$neSupportedNot equal
$gtSupportedGreater than
$gteSupportedGreater or equal
$ltSupportedLess than
$lteSupportedLess or equal
$inSupportedIn array
$ninSupportedNot in array
$existsSupportedField exists

Logical Operators

OperatorResultNotes
$andSupportedAND conditions
$orSupportedOR conditions

Data Type Support

TypeResultNotes
StringSupported
NumberSupportedInteger and float
BooleanSupported
ArraySupportedFor $in/$nin

Fetch Operations

FeatureResultNotes
Single IDSupported
Multiple IDsSupportedUp to 100/request
With namespaceSupported

Delete Operations

FeatureResultNotes
By IDSupportedSingle or multiple
By filterSupportedMetadata filter
Delete allSupportedClear namespace
By namespaceSupported

Update Operations

FeatureResultNotes
Update metadataSupported
Update valuesSupportedReplace vector
Sparse valuesSupportedHybrid vectors

Namespace Support

FeatureResultNotes
Create namespaceSupportedOn upsert
List namespacesSupportedIn stats
Query namespaceSupported
Delete namespaceSupported
FeatureResultNotes
Upsert sparseSupportedindices + values
Query sparseSupportedCombined scoring
Dense + sparseSupportedFull hybrid

Metadata

FeatureResultNotes
String fieldsSupported
Numeric fieldsSupportedInteger, float
Boolean fieldsSupported
Array fieldsSupportedFor filtering
Nested objectsPartialFirst level only
Max size40KBPer vector

Index Configuration

SettingResultNotes
DimensionSupportedUp to 20000
MetricSupportedcosine, euclidean, dotproduct
PodsSupportedPod-based scaling
ReplicasSupportedRedundancy
Pod typeSupportedPerformance tier

Statistics

MetricResultNotes
Total vectorsSupported
DimensionSupported
Index fullnessSupported
Namespace statsSupportedPer-namespace counts

SDK Compatibility

Python SDK

FeatureResultNotes
pinecone-clientSupportedv2.x, v3.x
UpsertSupported
QuerySupported
FetchSupported
DeleteSupported
UpdateSupported
Batch operationsSupported

Node.js SDK

FeatureResultNotes
@pinecone-database/pineconeSupportedv1.x, v2.x
All operationsSupported

Go SDK

FeatureResultNotes
pinecone-go-clientSupportedv0.x
All operationsSupported

REST API

Request Format

FormatResultNotes
JSONSupportedContent-Type: application/json
gRPCNot SupportedREST only

Authentication

MethodResultNotes
API KeySupportedHeader: Api-Key
EnvironmentSupportedPINECONE_API_KEY

Performance

MetricValueNotes
Query latency<10msFor top_k=10
Upsert throughput100K/secWith batching
Max vectors/index10M+Configurable
Max dimension20000Configurable

Known Limitations

Not Supported

FeatureReason
gRPC ProtocolREST API only
CollectionsUse namespaces
Pod type: s1Use standard pods
Serverless (AWS)On-premise only

Behavioral Differences

  1. Indexing: Uses HeliosDB’s HNSW implementation
  2. Sharding: Different sharding strategy
  3. Consistency: Strong consistency by default
  4. Cold start: No cold start delays

Related: README.md | CONFIGURATION.md | EXAMPLES.md