FrontierOR - Feasibility: leaderboard

Feasibility in percent: the share of large instances over all 180 tasks on which programs that LLMs generate in one shot for large-scale optimization problems from operations research papers, run on hidden large instances against an expert-verified Gurobi reference (FrontierOR, 2026) return a feasible solution within the time budget. Higher is better.

Metric: Feasibility times 100 (%): share of large instances over all 180 tasks on which the program returns a feasible solution within the time budget, one-shot generation of a complete optimization program from the natural-language problem and instance schemas (a few self-debug iterations allowed, tiny-instance sanity gate), run on the hidden large instances against an expert-verified Gurobi reference; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 4.662#113
2Gemini 3.1 Pro (Preview)61#99
3GPT-5.3 Codex60#124
4DeepSeek R142#482
5Grok 4.2028#301
6Qwen 3 Coder Plus26#720
7Llama 4 Maverick18#1110

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=frontieror-feasibility · How It Works · Data refreshed daily, snapshot 2026-11-21.