Bus Scheduling on NYC Manhattan (in-domain)
256.16Fuel ConsumptionTrafficClaw (8B)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TrafficClaw (8B)Model size=8B2026.04 | 256.16 | 437.17 | |
| AIDE2026.04 | 257.58 | 511.95 | |
| Qwen3-32BFramework=TrafficClaw2026.04 | 295.91 | 532.29 | |
| Qwen3-8BFramework=TrafficClaw2026.04 | 307.3 | 502.74 | |
| Classic Method2026.04 | 339.69 | 512.11 | |
| Qwen3-235BFramework=TrafficClaw2026.04 | 429.8 | 490.24 | |
| SUMO-MCP2026.04 | 702.45 | 504.87 | |
| TrafficGPT2026.04 | 749.69 | 498.83 | |
| o4-miniFramework=TrafficClaw2026.04 | 800.09 | 434.84 | |
| DeepSeek-V3.2Framework=TrafficClaw2026.04 | 927.15 | 419.42 | |
| Gemini-3.1-ProFramework=TrafficClaw2026.04 | 1,006.75 | 398.29 | |
| Kimi-K2.5Framework=TrafficClaw2026.04 | 1,195.44 | 391.98 | |
| MiniMax-M2.5Framework=TrafficClaw2026.04 | 1,476.15 | 479.56 |