Benchmark our solver against Google, or anyone.
The Elara-Cortex engine solves vehicle-routing and scheduling problems. The mathematics is solved, not guessed. Where optimality is provable, it matches the proven global optimum. Where it is not, it beats Google OR-Tools on the public benchmarks we ran, by a wider margin on the hardest problems. Don't take our word for it: register, send your own problem, and get back a solution certificate you can verify and compare.
Their solver guesses; ours is solved. Verified by the Z3 theorem prover: on instances small enough to prove the optimum, our engine reaches it 5 / 5. On the public CVRPLIB benchmark (100–512 stops) it averages 3.05% from the world's best-known solutions and beats Google OR-Tools on every instance. Full method, tables and verifiable certificates in the technical report.
The open challenge, three steps, your data
# 1 · get a key (7-day trial, no card) -> route.elara-cortex.com/developers
# 2 · POST your routing/scheduling instance, coordinates or a full distance matrix
curl -X POST https://route.elara-cortex.com/v1/solve -H "X-API-Key: $KEY" -d '{
"kind": "cvrp",
"coords": [[0,0],[12,5],[3,9], ...], # node 0 = depot
"demands": [0, 7, 4, ...],
"capacity": 100,
"budget_s": 5
}'
# 3 · get a SOLUTION CERTIFICATE back, routes + an independent verification block
{ "ok": true, "vehicles": 9, "cost": 27595,
"verified": { "coverage": true, "capacity": true, "cost_match": true },
"routes": [[...]], "solve_ms": 5002 }
# now run the SAME instance through Google OR-Tools, Gurobi, Hexaly or your own solver, compare like-for-like.
Supports capacitated routing (cvrp) and hard time
windows (vrptw) today. The returned routes are the certificate -
verifiable with the instance and arithmetic alone, revealing nothing about how they were found.
Why teams switch
Provable quality
Matches the Z3-certified optimum where provable; beats the strongest free solver everywhere else, published, reproducible, certificate-backed.
Milliseconds to replan
The same engine re-plans a disturbed 100-stop fleet in ~1 ms and re-routes 302x faster online or off, optimisation you can afford to run on every change, not once a day.
Every modality
Road, drone, sea and autonomous routing, plus distance matrices and multi-stop scheduling, one API, one key.
Your IP stays yours
You send the problem; we return the answer. Self-host keeps data in your tenancy. We never see your demand model.
Built for industry scheduling at scale
Routing and scheduling is the same mathematics across industries. The API serves it for:
| Sector | Problem the API solves |
|---|---|
| Logistics & distribution | Multi-vehicle delivery planning, dispatch, live re-routing, distance matrices |
| Field service & utilities | Technician scheduling with time windows, capacity and skills |
| Mining & heavy industry | Haul-fleet routing, asset tracking vs plan, dead-zone offline routing |
| E-commerce & retail | Same-day and next-day fleet planning, surge re-planning |
| Aviation | Aircraft routing, ground-fleet scheduling, disruption recovery, see the enterprise aviation API → |
Benchmark our solver against Google, or anyone.
Provably optimal where provable · beats OR-Tools on the public benchmarks we ran · 7-day free trial, no card.