Optimization Engine · in-app & as an API

Sequence thousands of stops into clean, drivable routes — in seconds

The same engine that powers Delivery OS dispatch is available on its own. Send us stops, vehicles and constraints; get back optimized, sequenced routes your drivers will actually follow.

1,000+
stops per solve
94 sec
typical solve time
41%
avg fewer miles
99.9%
API uptime SLA
No more spaghetti
Geographically clustered, non-overlapping territories drivers trust.
Time windows honored
Per-stop windows, service times, and driver shift limits respected.
Capacity aware
Volume, weight and package count constraints per vehicle.
Production-grade
Idempotent jobs, retries, and an uptime SLA you can build on.
How it works

Three calls from a pile of addresses to a finished plan

This is the free overview. The full request/response reference, constraint catalog and tuning guide live in the developer docs below.

01

Send your stops

Post addresses or coordinates with optional time windows, service durations, and package sizes. We geocode anything that needs it.

02

Define your fleet

Describe vehicles, capacities, start/end depots, shift windows and breaks. Mixed fleets and multi-depot are supported.

03

Get optimized routes

Receive sequenced routes per vehicle with ETAs, distance and drive time — ready to dispatch or render on a map.

Capabilities

A real VRP solver, not a nearest-neighbor toy

Time windows
Hard and soft windows per stop.
Capacity & volume
Weight, volume and unit constraints.
Mixed fleets
Different vehicle profiles & costs.
Multi-depot
Many start/end locations.
Shifts & breaks
Driver hours and mandated breaks.
Balanced routes
Even workloads across drivers.
Live re-optimize
Recompute as stops change.
Priorities & SLAs
Must-serve and penalty handling.
Benchmarks

Fast at scale, measured against unoptimized baselines

Problem size
Solve time
Distance saved
100 stops · 4 vehicles
3.1 sec
−38% miles
500 stops · 12 vehicles
28 sec
−42% miles
1,000 stops · 25 vehicles
94 sec
−41% miles
2,500 stops · 60 vehicles
4.5 min
−39% miles

Indicative figures on standard infrastructure; results vary by geography and constraint mix. Full methodology available to API customers.

Developer API

One endpoint. JSON in, optimized routes out.

A single authenticated POST kicks off an optimization job. Poll or subscribe to a webhook for the result. SDKs for Node, Python and Go ship with an API key.

  • Bearer-token auth, scoped per project
  • Synchronous for small jobs, async + webhook for large
  • Deterministic, idempotent job ids — safe to retry
  • Rate limits and usage metering built in
POST /v1/optimize
curl https://api.deliveryos.app/v1/optimize \
  -H "Authorization: Bearer $DOS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "vehicles": [
      { "id": "van-1", "capacity": 120,
        "start": [43.65, -79.38],
        "shift": ["08:00", "17:00"] }
    ],
    "stops": [
      { "id": "s1", "location": [43.70, -79.40],
        "size": 3, "window": ["09:00", "12:00"] },
      { "id": "s2", "location": [43.66, -79.35],
        "size": 1, "service": 120 }
    ]
  }'
200 OK
{
  "job_id": "opt_3kQ9…",
  "routes": [
    { "vehicle": "van-1",
      "distance_km": 40.2,
      "drive_time_min": 135,
      "sequence": ["s2", "s1"] }
  ],
  "unassigned": []
}
Documentation

Start reading for free. Unlock the full reference with a key.

The optimization engine is our product, so the deep reference is available to customers and trial keys. Here is what is open, and what unlocks when you request access.

Free to read
Quickstart
Authenticate and run your first optimization in 5 minutes.
Core concepts
Stops, vehicles, depots, jobs — the mental model, explained.
Sample request
The annotated POST /v1/optimize shown above.
Unlocks with an API key
Full constraint catalog
Every window, capacity, skill, priority and penalty field.
Async jobs & webhooks
Large-job lifecycle, polling, signatures, retries.
Tuning & objectives
Balancing miles vs. time vs. fairness; cost models.
SDK reference
Node, Python and Go clients with typed models.
Rate limits & metering
Quotas, burst behavior, and usage accounting.
Multi-depot recipes
Reference architectures for complex fleets.
Request a key to read the full docs

Trial keys include complete documentation and a sandbox quota. No credit card to start.

API pricing

Pay for what you optimize

Transparent, usage-based pricing — no per-seat tax on your dispatchers.

Trial
Free
Sandbox key
  • Up to 250 stops / month
  • Full documentation access
  • Community support
  • Node / Python / Go SDKs
Most popular
Scale
Usage-based
Pay per optimized stop
  • Volume tiered pricing
  • Async jobs + webhooks
  • 99.9% uptime SLA
  • Email & chat support
  • Higher rate limits
Enterprise
Custom
Dedicated & on-prem
  • Private deployment options
  • Custom constraints & objectives
  • Solution architect
  • Security review & DPA
  • Premium SLA
FAQ

Questions, answered

Can I use the optimizer without the full Delivery OS platform?+

Yes. The optimization engine is available as a standalone API. Many customers embed it in their own dispatch tooling while others use the full platform.

How is it priced?+

Usage-based, billed per optimized stop with volume tiers. There are no per-seat fees. A free trial key includes a monthly sandbox quota.

What constraints are supported?+

Time windows, service durations, vehicle capacity and volume, mixed fleets, multi-depot, shifts and breaks, priorities and SLAs. The full catalog is in the gated docs.

How large a problem can it handle?+

Thousands of stops across dozens of vehicles. Small jobs return synchronously; large jobs run async with a webhook callback.

Why are some docs gated?+

The engine is our core IP. We keep the quickstart and concepts open so you can evaluate the fit, and unlock the full reference the moment you request a key.

Request access to the optimization API

Tell us about your fleet and volumes and we'll set you up with a trial key, full documentation, and a sandbox quota.

Prefer email? sales@deliveryos.app