Skip to content

GeoLibre vs. other GIS platforms

How GeoLibre compares to the desktop GIS, cloud GIS, and web-mapping tools people most often ask about: QGIS, ArcGIS Pro, ArcGIS Online, CARTO, Felt, and kepler.gl.

GeoLibre is not trying to replace any of these outright. It occupies a spot none of them quite fills: a free and open-source GIS that runs in a browser tab with nothing installed, keeps your data on your own machine, and still ships real analysis, spatial SQL, cartography, and a project file — then packages the same app as a desktop install, an Android app, and a Jupyter widget.

How to read this

Comparisons like this age quickly and are written by an interested party. The GeoLibre column reflects what ships today; the other columns describe each product's mainstream, out-of-the-box behavior as of August 2026, not what is reachable with every add-on, extension, or third-party plugin. Where a capability arrives through a separate product (QField for QGIS in the field, ArcGIS Field Maps for ArcGIS), the table says so rather than crediting it to the base product. Found something wrong or out of date? Please open an issue — corrections are welcome.

At a glance

GeoLibre QGIS ArcGIS Pro ArcGIS Online CARTO Felt kepler.gl
License MIT, open source GPL-2.0+, open source Proprietary Proprietary Proprietary Proprietary MIT, open source
Cost Free Free Paid subscription Paid subscription (credit-metered analysis) Paid subscription Paid subscription Free
Browser Full app, nothing to install No official browser build No Yes (the product is the browser app) Yes Yes (the authoring product) Yes
Desktop Windows, macOS, Linux (Tauri) Windows, macOS, Linux Windows only — — — —
Mobile Native Android app; responsive touch layout Via QField / Mergin Maps (separate apps) — Via ArcGIS Field Maps (separate app) — Felt Field App for iOS and Android (separate app) Responsive web
In Jupyter Full app as an anywidget, two-way sync Via qgis bindings, not the UI Notebooks drive arcpy, not the UI ArcGIS API for Python Via pydeck-carto — Yes (widget)
Works offline Yes — PWA install, offline area download, desktop build Yes Yes Limited (Field Maps offline areas) No — connected platform by design Field App offline areas, syncing on reconnect (higher plans) Client-side, but assets are hosted
Where your data lives Your device — processed client-side in the browser session Your device Your device / your enterprise geodatabase Vendor cloud Your own cloud data warehouse — no CARTO-side storage (except cache) or sync; imports write to a warehouse you own Vendor cloud, or a single-tenant instance in your own AWS account (Enterprise) Your browser
Project file .geolibre.json (open, documented) .qgs / .qgz (open) .aprx (proprietary) Web map JSON (hosted) Map/Workflow JSON via CLI and MCP Hosted map (no local file) Exportable map config JSON

Data and formats

GeoLibre QGIS ArcGIS Pro ArcGIS Online CARTO Felt kepler.gl
Format breadth Wide — DuckDB-WASM Spatial plus in-house readers Widest — everything GDAL/OGR reads Very wide, plus native Esri formats Common upload formats Common upload formats + RaQuet for raster data Common upload formats CSV, GeoJSON, Arrow/Parquet
Cloud-native vector GeoParquet, FlatGeobuf, PMTiles, streamed over HTTP range requests GeoParquet, FlatGeobuf, PMTiles (recent GDAL) GeoParquet (recent), no native PMTiles — GeoParquet Some cloud sources on higher plans Arrow/Parquet
Cloud-native raster COG, Zarr, Cloud-Optimized NetCDF/HDF, kerchunk, MBTiles COG, Zarr (GDAL's built-in driver) COG; Zarr as a native multidimensional raster type Hosted imagery layers COG and GeoTIFF, loaded into warehouse tables (RaQuet spec) GeoTIFF upload —
STAC Built-in catalog browser + STAC Index discovery Built-in STAC connections (Browser panel and Data Source Manager) Built-in STAC connections and the Explore STAC pane Living Atlas (not STAC) — — —
OGC services XYZ; WMS/WFS/WMTS discovered via GetCapabilities; OGC API Features/Tiles from a landing page, collection, or items URL Full OGC support Full OGC support Yes Limited Limited XYZ / vector tiles
Esri services and geodatabases ArcGIS FeatureServer, VectorTileServer, I3S, plus local .gdb File Geodatabase folders (desktop) Yes Native Hosted feature, tile, imagery, and scene layers One-way ArcGIS Online / Portal migration tooling only Some ArcGIS layers —
Databases PostGIS browsing, DuckDB, in-browser PGlite/PostGIS PostGIS, SpatiaLite, Oracle, SQL Server, SAP HANA Enterprise geodatabases (SDE) Hosted only The core product — BigQuery, Snowflake, Redshift, Databricks, PostgreSQL/PostGIS, Oracle Spatial, all queried live PostGIS / Snowflake connections (higher plans) —
3D and point clouds LiDAR, 3D Tiles, I3S, Gaussian splats, glTF/GLB, Cesium globe pane Point clouds, 3D map view, tiled scene layers Deepest 3D — scenes, I3S, mesh, LAS Scene Viewer 3D map view, extrusion, Google Photorealistic 3D Tiles basemap. No point clouds or LiDAR — Extrusions and hexbins only
Planetary basemaps Moon, Mars, Mercury, Venus, Galilean moons, Titan, Pluto, Charon — with a per-project ellipsoid driving measurements Via plugins Limited — — — —

Analysis and processing

GeoLibre QGIS ArcGIS Pro ArcGIS Online CARTO Felt kepler.gl
Geoprocessing tools 1,000+ (Whitebox suite + GeoLibre's own), running in the browser on WebAssembly 1,000+ (native, GDAL, GRASS, SAGA) Most mature and complete, plus paid extensions A useful subset, credit-metered 200+ Workflows components and 180+ Analytics Toolbox SQL functions, executed as SQL in your warehouse A small set of common tools No tool catalog — analysis goes through SQL and the AI assistant
Where analysis runs Your browser (WASM), no server required; optional Python sidecar on desktop Your machine Your machine, or a server/portal Vendor cloud Your cloud data warehouse — queries are pushed down live, with repeated identical queries served from CARTO-managed cache Vendor cloud Your browser
Vector analysis Buffer, overlay, dissolve, joins, selection, topology checks — Turf.js, or GeoPandas via Pyodide/sidecar Comprehensive Comprehensive Common tools Buffer, overlay, spatial join, clip, Voronoi/Delaunay, KNN, trade areas — as warehouse SQL Common tools Spatial joins via SQL or the AI assistant
Raster analysis Hillshade, slope, contour, zonal/focal stats, raster calculator, reclassify, mosaic — rasterio sidecar with a browser fallback Comprehensive (GDAL/GRASS/SAGA) Comprehensive; Spatial Analyst extension Limited Narrow — zonal statistics and band value extraction — —
Spatial SQL DuckDB Spatial, PGlite/PostGIS, and Apache Sedona — all in the browser Virtual layers, DB Manager, PostGIS connections SQL against enterprise geodatabases — The native interface — warehouse SQL everywhere: query sources, SQL parameters, Workflows, SQL API SQL on connected sources (higher plans) DuckDB SQL Data Explorer over loaded data and remote URLs
Model / batch chaining Batch runner with model and pipeline chaining Graphical Model Designer ModelBuilder — Workflows — visual DAG builder with scheduling, version history, and SQL export — —
Spatial statistics Toolbox including Emerging Hot Spot Analysis Via plugins and processing providers Full Spatial Statistics toolbox Some Deep — Getis-Ord, Moran's I, space-time hotspots, GWR, kriging/IDW, composite scores, twin areas — —
Network analysis Isochrones, service areas, OD cost matrices, routing QGIS Network Analysis, plugins (ORS, pgRouting) Network Analyst extension Routing services (credits) Isolines, routes, and routing matrices via Location Data Services (credit-metered) — —
AI / ML AI Segmentation (SamGeo/SAM 3), in-browser ONNX/YOLO object detection Via plugins arcgis.learn deep-learning toolset Pretrained models (credits) BigQuery ML and Snowflake ML components; embeddings and geospatial foundation models — —
Processing history Every run listed, re-runnable, with copyable Python History panel Geoprocessing history — Workflow version history and run logs; org-wide activity/audit log (Enterprise) — —

Cartography, styling, and layout

GeoLibre QGIS ArcGIS Pro ArcGIS Online CARTO Felt kepler.gl
Renderers Single, categorized, graduated, rule-based, expression, heatmap, cluster, proportional symbols, diagrams Richest — plus every symbol layer type Very rich Smart-mapping styles Point, line, polygon, grid, H3, raster — plus heatmap and cluster aggregation; quantile/quantize/log/custom scales Simple, well-designed defaults Visualization-oriented
Data-defined everything Expression Builder wired into filters, labels, styling, field calculation, selection Yes, pervasive Attribute-driven symbology (Arcade) Arcade Attribute-driven styling per channel; custom SQL aggregation expressions in place of an expression language — —
Labeling Data-defined engine with placement, offset, rotation, wrap, dedup Best-in-class placement engine Maplex label engine Basic Basic Basic Basic
Print layout Print Layout composer with legend, scale, title block, atlas / map series, PNG and PDF Full print composer with atlas Full layout view with map series Basic print / export Export to high-res image and PDF Export to image and PDF Image export
Style interchange Imports and exports OGC SLD, QGIS QML, and Mapbox GL JSON QML, SLD .lyrx, limited SLD — — — —
Project import Reads QGIS .qgs/.qgz and ArcGIS Pro .aprx/.mapx Reads its own Reads its own (and .mxd) — — — —
Story maps Built-in story map builder with presenter view and standalone HTML export Via plugins — ArcGIS StoryMaps (separate product) Built-in skill for storymaps app development Shareable maps, not chaptered stories —
Dashboards Dashboard panel of chart and indicator widgets with cross-filtering Via plugins — ArcGIS Dashboards (separate product) Core to CARTO Builder — formula, category, pie, histogram, range, time series, and table widgets with cross-filtering Dashboard components — statistics, bar, histogram, time series Charts and filters in-map

Automation, extensibility, and sharing

GeoLibre QGIS ArcGIS Pro ArcGIS Online CARTO Felt kepler.gl
Scripting Python Console, geolibre Python package, docked Jupyter notebook beside the map PyQGIS, qgis_process headless CLI arcpy, notebooks ArcGIS API for Python / JavaScript CARTO + deck.gl + CARTO CLI + REST APIs REST API + Python client JS library
Notebook integration The whole app embeds as a Jupyter anywidget with two-way project sync Bindings, not the UI Notebooks in Pro drive arcpy Hosted notebooks pydeck-carto widget in Jupyter or Colab; Analytics Toolbox SQL from any warehouse notebook — Widget
Plugin ecosystem TypeScript plugin API, built-ins, zip installs, bundled drop-ins Largest — thousands of Python plugins .NET add-ins, Python toolboxes Configurable app templates Workflows extension packages None (API only) Fork the library
AI assistant Natural-language assistant that turns plain English into auditable, undoable operations; bring your own API key Via plugins Copilot features Arcade assistant (beta) and the AI assistants family AI Agents on a map — semantic model, configurable tools, bring your own LLM — plus a hosted MCP server, CLI, and agent skills AI-assisted map making Yes — OpenAI, Gemini, DeepSeek, or local Ollama; bring your own key
Real-time collaboration Yes (MVP) — hosted relay, optionally self-hosted — plus anchored review comments No (Mergin Maps for sync) No Shared editing of hosted layers Asynchronous plus real-time map-anchored comments Best-in-class multiplayer editing and comments —
Embedding maponly and layout=viewer URL modes, a versioned postMessage API, and the typed @geolibre/embed npm client QGIS Server / QGIS2Web export — Embeddable web apps Embeddable private maps with URL parameters and a postMessage API Embeddable maps Embeddable
Self-hosting Docker image, or serve the static build anywhere QGIS Server ArcGIS Enterprise (paid) — Yes — Docker on a single VM or Kubernetes, on any vendor cloud or on-premises Single-tenant AWS VPC, Felt-maintained (Enterprise) Static build
Standalone export Whole project to one offline HTML file, no server Via qgis2web plugin — — Data export to CSV, GeoJSON, GeoPackage, GeoParquet, KML, and Shapefile — HTML export

Where each one is the right choice

Choose GeoLibre when you want a real GIS in a browser tab with nothing to install, your data has to stay on your machine, you work with cloud-native formats (COG, GeoParquet, PMTiles, Zarr, STAC), you want the same app on desktop, Android, and inside a notebook, or you are embedding a map workspace into your own product and want an open, documented project format and license.

Choose QGIS when you need the widest possible format support through GDAL/OGR, the deepest desktop cartography and label placement, a mature plugin for a niche task, or heavy local processing over datasets larger than a browser can hold.

Choose ArcGIS Pro when you are already in the Esri ecosystem, need enterprise geodatabase editing and versioning, the deepest 3D and imagery workflows, or a specific specialized toolbox with organizational support behind it.

Choose ArcGIS Online when the priority is hosted layers, organizational sharing and permissions, and the surrounding Esri app family (Dashboards, StoryMaps, Field Maps, Experience Builder).

Choose CARTO when your data already lives in a cloud data warehouse and you need it to stay there. It suits teams that want AI Agents, visual analytics pipelines, dashboards and/or custom apps for large-scale data that non-analysts can use.

Choose Felt when collaborative, low-friction map making with a team is the whole job, and polished multiplayer editing and field data collection matter more than analysis depth. Enterprise plans can run a single-tenant instance in your own AWS account if data residency is the blocker.

Choose kepler.gl when you want fast, beautiful exploratory visualization of large point and trip datasets, and DuckDB SQL plus its AI assistant cover the analysis you need — rather than a geoprocessing toolbox, cartographic output, or a portable project file.

They also compose. GeoLibre reads QGIS and ArcGIS Pro projects and exchanges symbology as SLD, QML, and Mapbox GL JSON, so it is reasonable to author in QGIS or Pro and publish or embed with GeoLibre.

Where GeoLibre is not the strongest option

Stated plainly, so the table above is worth trusting:

  • Very large local datasets. Browser memory and WASM are real limits. Client-side vector tiling and streaming push the ceiling up, but a multi-gigabyte local processing job still belongs on a desktop GIS. The desktop build's Python sidecar helps; QGIS or ArcGIS Pro helps more.
  • Format breadth. GDAL/OGR reads more than DuckDB-WASM Spatial does. Some formats — notably File Geodatabase — only work in the desktop build.
  • Depth of specialized toolboxes. ArcGIS Pro's Spatial Analyst, Geostatistical Analyst, and Network Analyst, and QGIS's twenty years of plugins, cover workflows GeoLibre has no equivalent for.
  • Enterprise data management. Versioned editing, replication, and enterprise geodatabase administration are outside GeoLibre's scope.
  • Maturity. GeoLibre is stable and in active development, but it is far younger than QGIS or ArcGIS. Some capabilities listed as shipping are recent — see Recently added.

See also