The mathematical optimization solver
for every problem class.
10× simpler, 10× faster. One API for every problem class: LP, QP, MILP, MINLP, QUBO, PUBO and NLP today, more on the way. No auth, no email. Free for a limited time.
import numpy as np
from ortools.math_opt.python import mathopt
from quicopt import Client
# N binary variables, QUBO matrix Q — given
model = mathopt.Model()
x = np.array([model.add_binary_variable() for _ in range(N)])
model.minimize(x @ Q @ x)
client = Client("https://try.quicoptapi.pgi.fz-juelich.de")
result = client.solve(model) # one call — no signup, no license file
result.objective # minimized xᵀQxWhere Quicopt wins.
LPs, MILPs, and convex QPs scale to millions of variables. The wall isn’t size, it’s structure: implicit Hessians, higher-order non-convex objectives, costs from a simulator. Quicopt handles them, through one simple pip install.
Implicit or dense Hessians
Smooth NLPs where Newton-class solvers stall, because the KKT system is too dense to factor, fills in catastrophically, or is only reachable through a simulator or learned model. Quicopt is purely first-order: gradients only, no Hessian, no factorization.
Higher-order, non-convex objectives
Degree-3+ polynomials and non-smooth logic don’t fit MILP/MIQP solvers natively. They require either massive auxiliary-variable reformulations or piecewise approximations that quietly change the problem. Quicopt solves the original objective directly.
Black-box objectives
When cost comes from a simulator, digital twin, or ERP model rather than a formula, gradients and branch-and-bound proofs don’t apply at all. Quicopt works from input–output evaluations.
10× simpler to use
pip install, one API: no auth, no email, no OR team, no enterprise procurement. First solve in minutes.
Your problem. Our solver. Your result.
How Quicopt works
Problem modeling
Your optimization problem is translated into a solver-ready model. Our team handles the modeling.
Solving
Our algorithms run on standard hardware. Full performance today.
Integration & Output
Results via REST API, CSV or directly into your ERP. No black box. Full traceability.
Self-Serve & Scale
pip install and start free. No auth, no email for a limited time. Pay only for what you solve, from laptop to production.
Where others give up, Quicopt starts.
“Quicopt lets our customers work in a whole new way. That’s worth a great deal.”
Where Quicopt earns its place.
Four application areas where the structural advantages above translate directly into faster, sharper solutions.
Electronics & EMS
Multi-supplier BOM with thousands of parts and volume-tier pricing. Quicopt's domain-specific presolve exploits the natural structure of sourcing problems (BOM hierarchy, demand independence, supplier overlap) to shrink the search space dramatically before solving. On our active pilot: 70× faster than CP-SAT.
Power systems
AC optimal power flow is non-convex by physics: sinusoidal power-flow equations break QP relaxations. Grid topology optimization adds binary switch decisions, making the problem MINLP even in the DC linearization. Quicopt handles both natively, with no convex relaxation.
Process & metals
Blending, pooling, and recipe optimization with bilinear or higher-order terms: concentration × flow, yield × throughput, alloy composition. Exact global solvers (BARON, SCIP, Couenne) prove optimality on small instances but scale poorly; Quicopt delivers high-quality primal solutions on full-plant models in seconds.
QUBO, PUBO & HUBO
Binary optimization of arbitrary polynomial degree: the problem class quantum computers are built for. Quicopt solves QUBO, PUBO and HUBO directly: no reduction of higher-order terms to quadratic, no auxiliary-variable blow-up. A classical alternative to a quantum computer, and it runs on a laptop.
Benchmarks
Head-to-head results on hard, public problems: measured, reproducible, and run on standard hardware.
Build with Quicopt
A free developer API, no signup, no key management.
pip install quicopt, build a standard Pyomo or OR-Tools MathOpt model in Python, and hand it to a single solve() call. Your first request sets up a free API key automatically. Docs, runnable examples, and the full client reference live in the Developer Hub.
- ✓ No signup, your first call sets up a key
- ✓ pip install quicopt
- ✓ Model in Pyomo or OR-Tools MathOpt
Scientific foundation. Commercial execution.
- 1Optimization algorithms Hyper-efficient heuristics that run on standard hardware.
- 2Hardware-agnostic framework One model, no reformulation. Runs on CPU and GPU.
- 3Strength at high complexity Scales with thousands of variables where established solvers need multi-hour runtimes.
- 4Research base PGI-12 / FZJ Developed at the Institute for Quantum Computing Analytics (PGI-12), Forschungszentrum Jülich.
- 5Peer-reviewed Methodology published in PRX Quantum, Physical Review A and further refereed journals.
Our Origin
Built on decades of research excellence in mathematical optimization.
Quicopt is a spin-off from Forschungszentrum Jülich, one of Europe’s largest interdisciplinary research centers. Backed by Helmholtz Enterprise, we translate cutting-edge research into industrial-grade optimization solutions.
Our team combines deep expertise in quantum computing, mathematical optimization, and high-performance computing. We deliver algorithms that are not just theoretically sound, but proven in real-world applications.
Backed by
See Quicopt on your own problem.
No generic scenario. We compute with your real data.
Request Live Demo
Whether you have a specific optimization problem or want to explore what’s possible, reach out.