Function Calling on BFCL
4.2Energy (Wh)Benign
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BenignTarget Model=Mistral-L2026.01 | 4.2 | 347.2 | 20 | |
| BenignTarget Model=L-DeepSeek2026.01 | 5.2 | 295.9 | 40 | |
| BenignTarget Model=Llama-3.32026.01 | 5.4 | 279.4 | 30 | |
| BenignTarget Model=GLM-4.52026.01 | 5.5 | 344.6 | 30 | |
| OverthinkTarget Model=Llama-3.32026.01 | 7.6 | 319.3 | 60 | |
| BenignTarget Model=Seed-32B2026.01 | 9.6 | 287.4 | 90 | |
| BenignTarget Model=Qwen-32026.01 | 13.9 | 273.3 | 40 | |
| OverthinkTarget Model=L-DeepSeek2026.01 | 21.9 | 324.2 | 800 | |
| OverthinkTarget Model=Mistral-L2026.01 | 55.9 | 307.4 | 320 | |
| OverthinkTarget Model=Qwen-32026.01 | 89.9 | 348.1 | 630 | |
| OverthinkTarget Model=Seed-32B2026.01 | 153.9 | 310.1 | 420 | |
| OverthinkTarget Model=GLM-4.52026.01 | 202.8 | 382.8 | 150 | |
| MCTS-based tool-layer DoS attackTarget Model=Qwen-32026.01 | 1,383.3 | 552.8 | 4,490 | |
| MCTS-based tool-layer DoS attackTarget Model=GLM-4.52026.01 | 1,649.3 | 495.3 | 3,520 | |
| MCTS-based tool-layer DoS attackTarget Model=L-DeepSeek2026.01 | 1,976.3 | 530.2 | 6,760 | |
| MCTS-based tool-layer DoS attackTarget Model=Seed-32B2026.01 | 2,105.4 | 576.4 | 4,870 | |
| MCTS-based tool-layer DoS attackTarget Model=Mistral-L2026.01 | 2,269.6 | 582.8 | 7,210 | |
| MCTS-based tool-layer DoS attackTarget Model=Llama-3.32026.01 | 2,448.5 | 526.7 | 6,610 |