Supply Chain Management on Beer Game 20-week, 2-2-2 lead times
1,063Total CostsLlama 4 Maverick 17B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Llama 4 Maverick 17BScenario=Reasoning + Orchestrator + Policy, Technique=demand+budget2026.05 | 1,063 | 33.15 | |
| Llama 4 Maverick 17BScenario=Reasoning + Policy, Technique=budget2026.05 | 1,235 | 38.51 | |
| Llama 4 Maverick 17BScenario=Reasoning + Orchestrator, Technique=demand2026.05 | 1,559 | 48.61 | |
| GPT-5 miniScenario=Reasoning + Orchestrator + Policy, Technique=demand+budget2026.05 | 1,596 | 49.77 | |
| GPT-5 miniScenario=Reasoning + Policy, Technique=budget2026.05 | 1,608 | 50.14 | |
| Llama 4 Maverick 17BScenario=Reasoning + Prompt, Technique=prompt2026.05 | 1,758 | 54.82 | |
| GPT-5 miniScenario=Reasoning + Orchestrator, Technique=demand2026.05 | 1,765 | 55.04 | |
| Llama 4 Maverick 17BScenario=Reasoning Model, Technique=default2026.05 | 2,080 | 64.86 | |
| GPT-5 miniScenario=Reasoning Model, Technique=default2026.05 | 2,142 | 66.79 | |
| GPT-4o miniScenario=Non-Reasoning + Orchestrator, Technique=demand+volatility2026.05 | 2,171 | 67.7 | |
| GPT-5 miniScenario=Reasoning + Prompt, Technique=prompt2026.05 | 2,620 | 81.7 | |
| Human TeamsScenario=Human2026.05 | 3,207 | 100 | |
| GPT-4o miniScenario=Non-Reasoning + Prompt, Technique=prompt2026.05 | 3,956 | 123.36 | |
| GPT-4o miniScenario=Non-Reasoning + Policy, Technique=budget2026.05 | 4,351 | 135.67 | |
| GPT-4o miniScenario=Non-Reasoning + Orchestrator, Technique=demand2026.05 | 4,383 | 136.67 | |
| GPT-4.1 miniScenario=Non-Reasoning + Prompt, Technique=prompt2026.05 | 4,734 | 147.61 | |
| GPT-4o miniScenario=Non-Reasoning Model, Technique=default2026.05 | 7,093 | 221.17 | |
| GPT-4.1 miniScenario=Non-Reasoning Model, Technique=default2026.05 | 7,093 | 221.17 | |
| Llama 3.3 70BScenario=Non-Reasoning Model, Technique=default2026.05 | 11,743 | 366.17 |