Ramp Metering on NYC Queens (out-of-domain)
95.09Travel TimeKimi-K2.5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Kimi-K2.5Framework=TrafficClaw2026.04 | 95.09 | 13.23 | |
| TrafficClaw (8B)Model size=8B2026.04 | 100.51 | 14.16 | |
| Qwen3-235BFramework=TrafficClaw2026.04 | 103.13 | 12.54 | |
| Qwen3-8BFramework=TrafficClaw2026.04 | 108.29 | 12.22 | |
| o4-miniFramework=TrafficClaw2026.04 | 126.31 | 12.62 | |
| Qwen3-32BFramework=TrafficClaw2026.04 | 141.16 | 12.38 | |
| Classic Method2026.04 | 148.21 | 11.46 | |
| AIDE2026.04 | 149.85 | 12.58 | |
| MiniMax-M2.5Framework=TrafficClaw2026.04 | 153.07 | 11.75 | |
| SUMO-MCP2026.04 | 163.27 | 11.91 | |
| Gemini-3.1-ProFramework=TrafficClaw2026.04 | 165.81 | 12.08 | |
| DeepSeek-V3.2Framework=TrafficClaw2026.04 | 206.91 | 12.97 | |
| TrafficGPT2026.04 | 296.66 | 12.73 |