Operations Research Problem Solving on NL4OPT
96.7Pass@1 AccuracyEvoOptiGraph
Evaluation Results
| Method | Links | |
|---|---|---|
| EvoOptiGraphParams=8B, Category=Fine-tuned LLMs2026.06 | 96.7 | |
| OptMATH (origin)Params=32B, Category=Fine-tuned LLMs2026.06 | 95.9 | |
| LLMOPT (origin)Params=14B, Category=Fine-tuned LLMs2026.06 | 80.3 | |
| DeepSeek-V3Params=671B, Category=Zero-shot LLMs2026.06 | 79.9 | |
| Qwen2.5-72B-InstParams=72B, Category=Zero-shot LLMs2026.06 | 79.9 | |
| OptiMUS-v0.3Params=Closed, Category=Agentic Methods2026.06 | 79.8 | |
| Qwen3-32BParams=32B, Category=Zero-shot LLMs2026.06 | 76.1 | |
| ORLMParams=8B, Category=Fine-tuned LLMs2026.06 | 73.8 | |
| GPT-4oParams=Closed, Category=Zero-shot LLMs2026.06 | 73.4 | |
| CoEParams=Closed, Category=Agentic Methods2026.06 | 66.7 | |
| CoTParams=Closed, Category=Agentic Methods2026.06 | 62.2 |