Comparison Wherobots vs. Apache Sedona: three paths to spatial at enterprise scale Most enterprise spatial work starts on Apache Sedona. As workloads grow into joins at scale, raster processing, nearest-neighbor queries, and production SLAs, teams are best served by the fully managed and most capable Physical AI compute cloud, Wherobots, or can also use a drop-in enterprise runtime for Apache Sedona. Talk to sales Start for free Wherobots Cloud The context engine for physical world data, fully managed, and most capable. Apache Sedona The open-source distributed spatial engine where most teams start. WherobotsDB BYOS A drop-in replacement for Apache Sedona with enterprise support. Feature comparison Wherobots Cloud, WherobotsDB BYOS, and Apache Sedona side by side Capabilities grouped by area, from core spatial SQL to raster, Earth intelligence and perception, and spatial AI tools. Wherobots Cloud WherobotsDB BYOS Apache Sedona Wherobots Cloud, WherobotsDB BYOS, and Apache Sedona side by side Capability Option A Wherobots Cloud Fully managed spatial compute, RasterFlow, Havasu catalog, and Spatial AI tools. Option B WherobotsDB BYOS Bring Your Own Spark: a drop-in runtime for existing Sedona clusters. Option C Apache Sedona Open source, Apache 2.0. Where most teams start. Fit and economics Best for Maximum performance, Earth intelligence and perception, managed infrastructure and catalog, and spatial AI developer tools Enterprise workloads that must stay on self-managed Spark Teams starting with spatial analytics, proofs of concept, Spark experts comfortable tuning by hand, and less complex workloads not requiring enterprise support Cost model Pay as you go WherobotsDB license plus your compute Free (Apache 2.0) Infrastructure Included: Fully managed by Wherobots Customer managed (drop-in runtime) Customer managed Enterprise support Included: Dedicated SLAs Included: Dedicated SLAs Not included: Community only Spatial compute capabilities Spatial SQL functions Vector and raster functions in the function reference below Included: 389, including 31 exclusive to Wherobots Included: All of Apache Sedona plus Wherobots-exclusive functions Included: 358, the full Apache Sedona set Query optimization Join strategy, skew prevention, spatial acceleration Included: Automatic, plus purpose-built engine optimizations Included: Automatic join selection and dynamic optimization Not included: Manual tuning Spatial joins at scale SpatialBench SF-1000, 12 query patterns Included: Completes all 12 Included: Completes all 12 Partial: Completes 6 of 12 (Q1 to Q5, Q7) Approximate KNN Find the N nearest features to each record Included: Available, with optimized runtime Included: Available Not included: Exact comparison only, impractical at enterprise volumes Apache Sedona API compatibility Included: 100% compatible Included: 100% compatible Included: Native Raster and out-of-database processing Out-of-database raster Load pixels on demand from object storage or STAC Included: Full support with managed runtime optimizations Included: On-demand loading, caching, Cloud Optimized GeoTIFF Not included: Loads full files into executor memory Raster processing and map algebra Zonal statistics, raster algebra, raster to vector, multi-band Included: Full suite plus distributed raster tile generation Included: Full suite with spatial filter push-down Included: In-memory raster processing Earth intelligence and perception with RasterFlow Wherobots Cloud only Mosaics from satellite and aerial imagery Sentinel-2, NAIP, and custom sources over an area of interest Included: Public Preview Not included: Cloud only Not included Computer vision inference at planetary scale Semantic segmentation, regression, patch-based processing Included: Public Preview Not included: Cloud only Not included Built-in models SAM3, Fields of the World, Tile2Net, Meta canopy height, ChesapeakeRSC Included: Five built-in models plus bring your own Not included: Cloud only Not included Bring your own PyTorch model Included Not included: Cloud only Not included Vectorize model outputs Turn raster predictions into geometries for spatial analysis Included Not included: Cloud only Not included Analytics and location intelligence Geostatistics DBSCAN clustering, Getis-Ord Gi* hot spots, outlier detection Included: Built in, distributed, and tuned for scale Included: Available Included: Available Location intelligence Reverse geocoding with Overture Maps, isochrones, map matching Included: Included, plus PMTiles generation Partial: Optional add-on Not included Spatial AI developer tools Wherobots Cloud only Spatial AI Coding Assistant for VS Code Build notebooks, submit jobs, and manage workspaces from VS Code Included: All editions Not included: Cloud only Not included Wherobots MCP Server Natural-language data discovery and SQL generation in Claude, Cursor, and VS Code Copilot Included: Available Not included: Cloud only Not included Wherobots CLI Automate spatial workflows from the command line Included Not included: Cloud only Not included Performance Performance vs Apache Sedona Measured on identical infrastructure Included: About 3x faster on the Rust-native, Arrow-columnar runtime Included: About 2x faster Baseline Price-performance Included: 20 to 30% better price-performance Your infrastructure costs Baseline Sources: Wherobots documentation, Wherobots pricing, and SpatialBench results. Updated April 2026. Benchmark SpatialBench results: spatial joins at scale Both WherobotsDB options complete all 12 SpatialBench query patterns at scale factor 1000. Apache Sedona completes six. The six it does not complete are zone statistics within a search radius (Q6), radius joins (Q8), polygon overlap detection (Q9), zone statistics through a spatial join (Q10), cross-zone trip counting (Q11), and KNN joins (Q12). Wherobots Cloud WherobotsDB BYOS Apache Sedona SpatialBench results: spatial joins at scale Capability Wherobots Cloud WherobotsDB BYOS Apache Sedona SpatialBench SF-1000: 520 GB, 6 billion records, 12 query patterns Q1 to Q5, Q7 Filtering, aggregation, convex hull, and basic joins Included: Pass Included: Pass Included: Pass Q6 Zone statistics aggregated within a search radius Included: Pass Included: Pass Not included: Does not complete Q8 Radius join: count nearby features per building Included: Pass Included: Pass Not included: Does not complete Q9 Polygon overlap detection and IoU Included: Pass Included: Pass Not included: Does not complete Q10 Zone statistics through a spatial join Included: Pass Included: Pass Not included: Does not complete Q11 Cross-zone trip counting, two spatial joins per record Included: Pass Included: Pass Not included: Does not complete Q12 KNN join: five nearest neighbors per record Included: Pass Included: Pass Not included: Does not complete SpatialBench is the open benchmark for spatial query engines from the Apache Sedona project. Read about WherobotsDB on SpatialBench. How to choose Your Sedona code comes with you Wherobots Cloud is the fastest way to run spatial in production, with WherobotsDB, RasterFlow, and the Havasu Catalog in one managed platform. Apache Sedona code runs on it without changes, and teams that must stay on their own Spark clusters run the same code on WherobotsDB BYOS. Raster and Earth intelligence Out-of-database raster comes with both WherobotsDB options. RasterFlow, Earth intelligence and perception as a service that turns imagery into physical-world features, runs in Wherobots Cloud. Spatial AI developer tools The Spatial AI Coding Assistant, MCP Server, and CLI come with Wherobots Cloud and put spatial analysis within reach of the whole team. Infrastructure control and data residency If compute must stay on your own infrastructure for compliance, procurement, or architecture reasons, WherobotsDB BYOS adds enterprise spatial without a managed platform. Function reference Every spatial SQL function, by platform Search 389 Apache Sedona and WherobotsDB functions across 32 categories, filter by vector or raster, and see where each one runs. Need a function that is not listed? Wherobots adds custom ST_ and RS_ functions for customers on request. Search functions Data type Vector and raster Vector Raster Availability All platforms Wherobots only (not in Apache Sedona) Available in WherobotsDB BYOS Category All categories Address Functions Affine Transformations Aggregate Functions Aggregates Band Accessors Bounding Box Functions Clustering Functions Connected Components (CC) Predicates Geography Type Functions Geometry Accessors Geometry Constructors Geometry Editors Geometry Output / Export Geometry Processing Geometry Validation Linear Referencing Map Algebra Operators Measurement Functions Operators Output Overlay Functions Per-Pixel Math (Map Algebra) Pixel Functions Predicates Predicates / Topological Relationships Raster Accessors Raster Constructors Raster Geometry Functions Spatial Indexing (H3, S2, Bing Tiles, GeoHash) Spatial Reference System Spatial Statistics Tiles 389 functions Function Category Apache Sedona WherobotsDB BYOS Wherobots Cloud What it does ExpandAddress Address Functions Available Available Available libpostal address expansion ParseAddress Address Functions Available Available Available libpostal address parsing ST_GetReverseGeocodingLayers Address Functions Not available Available Available layers available to ST_ReverseGeocode ST_ReverseGeocode Address Functions Not available Available Available address for a geometry ST_Affine Affine Transformations Available Available Available general affine transformation ST_Rotate Affine Transformations Available Available Available rotate geometry; Wherobots supports custom origin ST_RotateX Affine Transformations Available Available Available rotate around X axis ST_RotateY Affine Transformations Available Available Available rotate around Y axis ST_Scale Affine Transformations Available Available Available scale geometry ST_ScaleGeom Affine Transformations Available Available Available scale using factor geometry ST_Translate Affine Transformations Available Available Available translate geometry by offsets ST_Analyze_Aggr Aggregate Functions Not available Not available Available aggregate geometry statistics ST_Collect_Agg Aggregate Functions Available Available Available collects without dissolving ST_Envelope_Agg Aggregate Functions Available Available Available return aggregate bounding envelope ST_Intersection_Agg Aggregate Functions Available Available Available aggregate polygon intersection ST_Union_Agg Aggregate Functions Available Available Available compute aggregate union ST_3DExtent Aggregate Functions Available Available Available 3D bounding box aggregate (Box3D) ST_Extent Aggregate Functions Available Available Available bounding box aggregate (Box2D) RS_Union_Aggr Aggregates Available Available Available stacks bands from multiple rasters RS_Band Band Accessors Available Available Available returns a new raster with one or more bands of an existing raster RS_BandIsNoData Band Accessors Available Available Available returns true if the band is filled with only nodata values RS_BandNoDataValue Band Accessors Available Available Available returns the no data value of the given band of the given raster RS_BandPixelType Band Accessors Available Available Available returns the pixel data type of a band as a string RS_Count Band Accessors Available Available Available returns the number of pixels in a given band RS_SummaryStats Band Accessors Available Available Available returns one summary statistic for a band RS_SummaryStatsAll Band Accessors Available Available Available returns count, sum, mean, standard deviation, min, and max for a band RS_ZonalStats Band Accessors Available Available Available returns one statistic computed over a zone geometry RS_ZonalStatsAll Band Accessors Available Available Available returns a struct of statistics computed over a zone geometry RS_ZonalStatsAllBands Band Accessors Not available Available Available zonal statistics for all bands in one call ST_BoundingDiagonal Bounding Box Functions Available Available Available diagonal linestring of bbox ST_Envelope Bounding Box Functions Available Available Available return 2D bounding box ST_Expand Bounding Box Functions Available Available Available expand bounding box ST_MMax Bounding Box Functions Available Available Available max M coordinate ST_MMin Bounding Box Functions Available Available Available min M coordinate ST_XMax Bounding Box Functions Available Available Available return max X coordinate ST_XMin Bounding Box Functions Available Available Available return min X coordinate ST_YMax Bounding Box Functions Available Available Available return max Y coordinate ST_YMin Bounding Box Functions Available Available Available return min Y coordinate ST_ZMax Bounding Box Functions Available Available Available return max Z coordinate ST_ZMin Bounding Box Functions Available Available Available return min Z coordinate ST_3DMakeBox Bounding Box Functions Available Available Available Box3D from two corner points ST_Box2D Bounding Box Functions Available Available Available planar bounding box as Box2D ST_Box3D Bounding Box Functions Available Available Available 3D bounding box as Box3D ST_GeomFromBox2D Bounding Box Functions Available Available Available Box2D to a rectangular polygon ST_MakeBox2D Bounding Box Functions Available Available Available Box2D from two corner points ST_DBSCAN Clustering Functions Available Available Available density-based spatial clustering ST_LocalOutlierFactor Clustering Functions Available Available Available LOF anomaly detection ST_ContainsCC Connected Components (CC) Predicates Not available Available Available spatial connected components via Contains ST_CoveredByCC Connected Components (CC) Predicates Not available Available Available spatial connected components via CoveredBy ST_CoversCC Connected Components (CC) Predicates Not available Available Available spatial connected components via Covers Available Not available Show all 389 functions No functions match these filters. Function names link to the Wherobots documentation. Wherobots vs. Apache Sedona FAQ What is the difference between Apache Sedona and Wherobots? Apache Sedona is the open-source spatial engine for Apache Spark, free under the Apache 2.0 license. Wherobots is the fully managed context engine for physical world data from the original creators of Sedona. It brings together WherobotsDB, the enterprise spatial runtime; RasterFlow, Earth intelligence and perception as a service that turns imagery into physical-world features; the Havasu Catalog of open datasets, RasterFlow models, notebooks, and apps next to your own data in Apache Iceberg; and Spatial AI developer tools, including the Spatial AI Coding Assistant for VS Code, the Wherobots MCP Server, and the Wherobots CLI. Apache Sedona code runs on Wherobots without changes. What is the difference between Apache Sedona and WherobotsDB? WherobotsDB is the enterprise spatial runtime from the original creators of Apache Sedona. It is 100% compatible with Sedona APIs and adds automatic query optimization, out-of-database raster, approximate KNN, and support SLAs. WherobotsDB runs on your own Spark clusters (BYOS) or fully managed in Wherobots Cloud. Is WherobotsDB compatible with Apache Sedona? Yes. WherobotsDB is 100% compatible with Apache Sedona APIs, so existing Sedona SQL and Python code runs without rewrites. Can I run WherobotsDB on my own Spark clusters? Yes. WherobotsDB Bring Your Own Spark (BYOS) is a drop-in runtime for existing Spark clusters, offered in the Enterprise edition. Data stays where it is and governance does not change. See pricing. Who builds Apache Sedona? The original creators of Apache Sedona are the founders of Wherobots. They continue to contribute to the project, which is maintained by the Apache Software Foundation community. What is SpatialBench? SpatialBench is an open benchmark for spatial query engines from the Apache Sedona project. It runs reproducible workloads such as point-in-polygon, range queries, spatial joins, and nearest-neighbor joins at multiple scale factors, so teams can compare engines on their own infrastructure. Does Apache Sedona support raster data? Yes, Apache Sedona processes rasters in memory. WherobotsDB adds out-of-database raster, which loads pixels on demand from object storage, and Wherobots Cloud adds RasterFlow for mosaics and computer vision inference over satellite and aerial imagery. Can Wherobots add a spatial function I need? Yes. Wherobots adds custom ST_ vector and RS_ raster functions for paying customers on request. Talk to us about your workload. See how your Sedona workloads run on Wherobots Bring a query or a dataset and we will compare Wherobots Cloud, Apache Sedona, and WherobotsDB BYOS on your workload. Schedule use case review
