FrontierOR: 180 Literature-Grounded Operations Research Tasks

180 self-contained operations research tasks benchmarked from academic literature. Each task includes a natural-language problem description, mathematical formulation, reference Gurobi implementation, test instances, and an automated feasibility checker. Created by SmartOR, the dataset was last updated on June 14, 2026.

Use Cases

• Benchmarking LLM performance on translating natural-language OR problems into executable code based on the provided problem descriptions.
• Evaluating the correctness of generated optimization code using the automated feasibility checker.
• Training or fine-tuning models for mathematical reasoning and code generation tasks based on the structured problem-solution pairs.
• Reproducing and comparing optimization algorithms using the provided reference Gurobi implementations and test instances.

Strengths

• Contains 180 distinct, literature-grounded tasks, providing a substantial benchmark scale.
• Each task is a self-contained reproducible unit with multiple components, including a reference implementation and automated checker.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

SmartOR

Collection Method

Packaged from academic literature, likely as described in the benchmark paper.

Freshness

Last updated 2026-06-14 11:13:13.