Bus Scheduling on NYC Queens (in-domain)
669.81Fuel ConsumptionClassic Method
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Classic Method2026.04 | 669.81 | 462.6 | |
| Qwen3-8BFramework=TrafficClaw2026.04 | 686.5 | 507.52 | |
| Qwen3-32BFramework=TrafficClaw2026.04 | 793.4 | 528.89 | |
| TrafficClaw (8B)Model size=8B2026.04 | 823.68 | 450.12 | |
| SUMO-MCP2026.04 | 903.04 | 471.71 | |
| Qwen3-235BFramework=TrafficClaw2026.04 | 980.78 | 494.84 | |
| o4-miniFramework=TrafficClaw2026.04 | 996.86 | 499.39 | |
| AIDE2026.04 | 1,042.66 | 449.21 | |
| TrafficGPT2026.04 | 1,170.44 | 564.85 | |
| DeepSeek-V3.2Framework=TrafficClaw2026.04 | 1,668.97 | 458.92 | |
| Gemini-3.1-ProFramework=TrafficClaw2026.04 | 1,880.37 | 441.01 | |
| MiniMax-M2.5Framework=TrafficClaw2026.04 | 2,282.35 | 468.99 | |
| Kimi-K2.5Framework=TrafficClaw2026.04 | 2,518.4 | 488.69 |