Maps & Geocoding APIs: Google Maps, Mapbox, HERE, Radar, MapTiler, TomTom, OpenStreetMap

⬅️ Backend & Data Overview

If you're building a SaaS that touches addresses, locations, routing, or maps, you need a maps + geocoding API. The naive approach: copy/paste a Google Maps embed and call it done. The structured approach: choose a provider based on use case (display vs geocoding vs routing vs autocomplete), pricing model (per-load vs per-request), and licensing (Google's strict ToS vs Mapbox's friendlier vector tiles vs OSM's free-but-DIY). The right pick depends on volume, geography (US-only vs global), and whether you need a polished consumer-grade map (Google) or a developer-friendly customizable map (Mapbox).

TL;DR Decision Matrix

The first decision is display vs no-display: do you need an interactive map UI, or just convert addresses to lat/lng (geocoding) and back (reverse geocoding)? The second decision is geography: US-only is much cheaper (Geocodio, Mapbox) than global. The third decision is license: Google has strict ToS (no caching geocode results); Mapbox / OSM are friendlier.

Decide What You Need First

Just a map display (the 30% case)

You want a map to show on a page (store locator, real estate listings, location pin). No address input.

Right tools:

• Google Maps — consumer-grade UX; widely recognized
• Mapbox GL JS — customizable; vector tiles; modern
• MapTiler — affordable Mapbox alternative
• Leaflet + OSM tiles — free + DIY

Geocoding only — address to lat/lng (the 30% case)

You take address strings and convert to coordinates (form normalization, sales territory mapping, distance calculation). No map UI.

Right tools:

• Google Geocoding API — high accuracy; strict ToS (no caching)
• Mapbox Geocoding — high accuracy; cacheable
• Geocodio — US/Canada-focused; bulk-friendly; cheap
• LocationIQ / Geoapify — OSM-based; cacheable; cheap

Address autocomplete (the 20% case)

You want a search box that suggests addresses as users type (checkout flow, signup, location finder).

Right tools:

• Google Places Autocomplete — best accuracy; expensive at scale
• Mapbox Search Box — good accuracy; cacheable; cheaper
• Algolia Places (deprecated, replaced by Algolia Search)
• Photon (OSM) — free; lower accuracy; self-host

Routing + directions (the 10% case)

You need to compute routes between A and B (last-mile delivery, sales-rep routing, transit time estimation).

Right tools:

• Google Directions API — driving / walking / transit; commute times
• Mapbox Directions — strong driving + cycling
• HERE Routing — fleet / commercial vehicle routing
• TomTom Routing — traffic-aware

Geofencing / location triggers (the 5% case)

You need backend events when users enter/exit zones (e.g., user pulled into store parking lot → send push).

Right tools:

• Radar — leader; mobile SDK + server-side
• Mapbox Movement — emerging
• Custom (PostGIS) — DIY for backend-only geofencing

Static map images (the 5% case)

You need server-rendered map images (email, PDF, social card thumbnails).

Right tools:

• Mapbox Static Images API — modern default
• Google Static Maps — established
• MapTiler Static Maps — affordable

Provider Deep-Dives

Google Maps Platform — the consumer default

The most-recognized maps brand. Premium quality + pricing.

Pricing in 2026:

• Maps embed: $7/1K loads (after $200/mo free credit)
• Geocoding: $5/1K requests
• Places Autocomplete: $17/1K sessions ($2.83/1K with sessions)
• Directions: $5-10/1K (driving) or higher (transit)
• Free $200/mo credit covers ~28K map loads OR ~40K geocoding requests

Features: maps embed, geocoding, autocomplete, directions, distance matrix, places details, Street View, traffic.

Why Google wins: highest accuracy globally; best autocomplete; consumer-recognized UX; deepest POI database.

Trade-offs: most expensive at scale; strict ToS (no caching geocode results, no using lat/lng outside Google Maps); requires API key + billing setup; vendor lock-in once integrated.

Pick if: B2C product where users expect Google's look; need autocomplete or POI search. Don't pick if: scaling B2B with millions of geocode calls (cost explodes).

Mapbox — customizable + developer-friendly

Mapbox has carved a niche as the "Google for developers." Vector tiles + custom styling + cacheable APIs.

Pricing in 2026:

• Maps GL JS: 50K loads/mo free, then $0.50-1/1K loads
• Geocoding (Search Box): 100K free, then $0.75/1K
• Directions: 100K free, then $1.50/1K
• Static images: $1/1K renders

Features: Mapbox GL JS (interactive maps), Vector + raster tiles, Studio (visual style editor), Geocoding (Search Box API), Directions, Static images, Movement (mobility data), Iso-routing.

Why Mapbox: customizable map styling (vector tiles); cacheable geocode results (vs Google's restrictive ToS); modern dev UX; React Native + Flutter SDKs.

Trade-offs: less recognized brand; slightly less accurate POI data than Google; pricing climbs at very high volume.

Pick if: B2B SaaS; want custom-styled maps; need cacheable geocoding. Don't pick if: need Google's specific UX or extensive POI database.

HERE — enterprise / logistics

Founded 1985 (was Nokia/Navteq). Strong enterprise + logistics player.

Pricing in 2026: free dev tier + custom enterprise pricing.

Features: maps, routing (truck/commercial-vehicle aware), geocoding, fleet management, traffic, isolines, autonomous-driving data (HERE HD Live Map).

Why HERE: enterprise-grade routing for trucking/logistics; private auto manufacturer data partnerships.

Pick if: logistics / fleet / commercial routing. Don't pick if: B2C / consumer maps.

Radar — geofencing + mobile SDK

Founded 2016. Mobile-first geofencing leader.

Pricing in 2026: 100K events/mo free; $0.50-1/1K events at scale.

Features: SDK (iOS/Android/web), geofences, places, trip tracking, geocoding (Google + own), insights.

Why Radar: best mobile SDK for geofences; battery-efficient; enterprise customers (Walgreens, T-Mobile).

Pick if: mobile app with location triggers (push when user nears store). Don't pick if: web-only or no geofencing need.

MapTiler — affordable Mapbox alternative

OSM-based map tiles + geocoding.

Pricing in 2026: 100K tiles/mo free; $25-500/mo paid.

Features: vector + raster tiles, geocoding, routing, static images, MapLibre GL JS support (open-source Mapbox alternative).

Why MapTiler: cheaper than Mapbox; OSS-friendly (MapLibre); good for cost-conscious SaaS.

Pick if: cost-conscious; OSS preference. Don't pick if: need bleeding-edge data quality.

TomTom — routing + traffic

Founded 1991. Strong traffic data + routing.

Pricing in 2026: limited free; enterprise contracts.

Features: maps, routing, traffic, navigation SDK.

Pick if: real-time traffic-aware routing critical. Don't pick if: simple maps suffice.

OpenStreetMap (OSM) — free + DIY

Community-built map data. Free; you host.

Pricing: free data; you pay hosting + bandwidth.

Tile rendering options:

• Self-host with OpenMapTiles stack (PostGIS + Tegola/MapTiler tooling)
• Use a managed OSM-based provider (Mapbox/MapTiler/Stadia)

Pick if: privacy-focused / cost-sensitive / OSS aligned. Don't pick if: need turnkey.

Geocodio — US/Canada-focused geocoding

Founded 2013. Excellent for bulk US/Canada geocoding.

Pricing in 2026: 2,500 req/day free; pay-per-use ($0.50/1K typical).

Features: forward + reverse geocoding, batch upload, congressional district / school district lookup.

Pick if: US-only geocoding; bulk operations. Don't pick if: international.

Nominatim / Pelias / Photon — OSS geocoders

Self-hosted OSS geocoders.

Pricing: free; you host (typically $20-200/mo VPS).

Pick if: privacy / cost-sensitive / DIY-comfortable. Don't pick if: need turnkey + accuracy.

What3Words — 3-word addressing

Maps every 3m × 3m square on earth to 3 words (///filed.most.cure).

Pricing: per-call API.

Pick if: last-mile delivery, hard-to-find locations, marine. Don't pick if: standard addressing works.

LocationIQ / Geoapify — OSM-based managed

Managed OSM geocoders + maps.

Pricing: $49-499/mo.

Pick if: cost-conscious; OSS values; cacheable APIs. Don't pick if: enterprise-procurement requires Google brand.

Esri ArcGIS — GIS / enterprise

Founded 1969. Enterprise GIS leader.

Pricing: enterprise; $$$$.

Pick if: enterprise GIS / federal / utilities. Don't pick if: standard SaaS maps.

What Maps APIs Won't Do

Buying a maps API doesn't:

1. Solve address normalization completely. Geocoding gets you 95-99% accuracy on well-formed addresses. Apartment numbers, PO boxes, rural routes, international formats break differently. Plan for fallback.
2. Avoid licensing constraints. Google's ToS forbids caching geocode results past 30 days, using results outside Google maps, etc. Mapbox is friendlier but still has limits. Read the ToS.
3. Replace user input validation. Users typo addresses. Autocomplete reduces typos but doesn't eliminate.
4. Scale costs predictably. A viral product can blow through $200/mo Google credit on day 1. Set budget alerts.
5. Make maps load fast on slow connections. Tiles + JS bundle = 1-3 MB initial. Plan loading state + low-bandwidth fallback.

The honest framing: maps APIs are infrastructure. Quality varies by region (Google is best in the US; HERE is best in Europe; OSM coverage varies). Test in your target geographies.

Cost Calculation Example

For a B2B SaaS with 100K geocoding requests/month + 50K map loads/month:

• Google: 100K × $5/1K = $500 + 50K × $7/1K = $350 → ~$650/mo (after free credit ~$650-200=$450/mo)
• Mapbox: 100K × $0.75/1K (after 100K free) = $0; 50K × $0.50/1K = $0 (after 50K free) → ~$0/mo
• MapTiler: similar to Mapbox; ~$25-100/mo
• Geocodio + Mapbox display: $0.50/1K × 100K = $50 + $0 maps → ~$50/mo

The Mapbox/MapTiler/Geocodio stack is dramatically cheaper than Google for B2B at scale. Most teams switch from Google to Mapbox once they hit the $200/mo free credit ceiling.

Pragmatic Stack Patterns

Pattern 1: Indie SaaS, MVP ($0-50/mo)

• Mapbox GL JS (50K free) for map display
• Mapbox Geocoding (100K free) for geocoding
• Total: $0/mo until viral

Pattern 2: B2B SaaS at scale ($50-500/mo)

• Mapbox for display + geocoding + directions
• Cache geocode results (Mapbox ToS allows it)
• Total: $50-500/mo

Pattern 3: US-only B2B ($30-300/mo)

• Geocodio for bulk geocoding
• Mapbox or Google for display
• Cheaper than full Google stack

Pattern 4: Mobile geofencing app ($100-1K/mo)

• Radar for geofences + mobile SDK
• Mapbox for display
• Total: $100-1K/mo

Pattern 5: Cost-sensitive / OSS ($0-100/mo)

• Self-hosted OSM (OpenMapTiles)
• Self-hosted Pelias / Nominatim for geocoding
• Total: $0-100/mo for hosting; engineering investment

Pattern 6: Enterprise logistics ($$$+)

• HERE or TomTom for routing
• Custom MSA pricing

Pattern 7: B2C consumer ($500-5K/mo)

• Google Maps for recognized UX
• Pay $500-5K/mo at scale

Decision Framework: Three Questions

1. What's the primary use case?

   • Map display only → Mapbox / Google / MapTiler
   • Geocoding only → Mapbox / Geocodio / LocationIQ / Google
   • Autocomplete → Google / Mapbox Search Box
   • Routing → Google / Mapbox / HERE / TomTom
   • Geofencing → Radar / Mapbox

2. What's your geography?

   • US-only → Geocodio + Mapbox cheapest
   • Global → Mapbox / Google
   • Enterprise EU/logistics → HERE / TomTom

3. What's your scale?

   • <100K req/mo → free tier (Mapbox / MapTiler)
   • 100K-10M → Mapbox / MapTiler paid (~$50-500/mo)
   • 10M+ → enterprise contracts (Google / Mapbox / HERE)

Verdict

For 50% of B2B SaaS in 2026: Mapbox. Best dev UX, customizable, cheaper than Google at scale, friendlier ToS.

For 25%: Google Maps. When users expect Google UX (real estate, store locator) or you need premium autocomplete.

For 10%: Geocodio + Mapbox. US-only B2B; bulk geocoding cheap, display via Mapbox.

For 5%: Radar. Mobile geofencing.

For 5%: MapTiler / Stadia / LocationIQ. Cost-sensitive; OSS-aligned.

For 5%: HERE / TomTom. Logistics / fleet / commercial routing.

The mistake to avoid: defaulting to Google without checking ToS. Caching geocode results is forbidden under Google's ToS — many SaaS teams accidentally violate. Mapbox / Geocodio explicitly allow caching.

The second mistake: not setting budget alerts. A viral product or runaway loop can spend $1K/day on Google Maps. Set hard caps + alerts.

The third mistake: picking by accuracy at home. The provider that's accurate in San Francisco may be terrible in Lagos or Mumbai. Test in target geographies before committing.

See Also

• IP Geolocation Providers — IP-to-location vs address-to-location
• API Integration — generic integration patterns
• Webhook Delivery Services — adjacent infrastructure
• SMS & Voice APIs — adjacent comms infra
• Notification Providers — push notifications for geofence events
• Database Providers — PostGIS for geo storage
• VibeWeek: Internationalization — locale + region context
• VibeWeek: Timezone Handling — adjacent localization
• LaunchWeek: Vertical SaaS Positioning — geography-specific positioning