By Hans Mittelmann (mittelmann at asu.edu)
Note that on top of the benchmarks a link to logfiles is given!
See this graphical tool for visualization of most of the results including a virtual best (ensemble)
Concorde, Hexaly and LKH on TSPLIB (7-28-2026)
CVRP problems from size 100 to 1 million (8-4-2026)
Concorde-TSP with different LP solvers (3-1-2025)
LPfeas Benchmark (find a PD feasible point; also for GPUs) (8-10-2026)
LPopt Benchmark (find optimal basic solution) (7-1-2026)
Large Network-LP Benchmark (commercial vs free) (7-1-2026)
MIPFEAS Benchmark - MIPLIB2017 (7-6-2026)
MILP Benchmark - MIPLIB2017 (7-7-2026)
MILP cases that are slightly pathological (7-14-2026)
Infeasibility Detection for MILP Problems (7-26-2026)
SQL problems from the 7th DIMACS Challenge (8-8-2002)
Several SDP codes on sparse and other SDP problems; also on GPUs (4-25-2026)
Infeasible SDP Benchmark (9-21-2025)
Large SOCP Benchmark (8-10-2026)
AMPL-NLP Benchmark (8-12-2026)
Non-commercial convex QP Benchmark (6-29-2026)
Non-Convex QUBO-QPLIB Benchmark (8-20-2026)
Binary Non-Convex QPLIB Benchmark (5-9-2025)
Discrete Non-Convex QPLIB Benchmark (non-binary) (5-12-2026)
Continuous Non-Convex QPLIB Benchmark (5-17-2026)
Convex Continuous QPLIB Benchmark (also on GPUs) (8-10-2026)
Convex Discrete QPLIB Benchmark (7-21-2026)