Throughput Efficiency on BFCL
5,093ThroughputBenign
Evaluation Results
| Method | Links | |
|---|---|---|
| BenignModel=Qwen-32026.01 | 5,093 | |
| OverthinkModel=Qwen-32026.01 | 4,561 | |
| BenignModel=Seed-32B2026.01 | 4,082 | |
| OverthinkModel=Seed-32B2026.01 | 4,078 | |
| BenignModel=L-DeepSeek2026.01 | 3,845 | |
| OverthinkModel=Llama-3.32026.01 | 3,822 | |
| OverthinkModel=L-DeepSeek2026.01 | 3,752 | |
| BenignModel=GLM-4.52026.01 | 3,734 | |
| BenignModel=Llama-3.32026.01 | 3,563 | |
| OverthinkModel=GLM-4.52026.01 | 3,185 | |
| BenignModel=Mistral-L2026.01 | 2,871 | |
| MCTS-based tool-layer DoS attackModel=GLM-4.52026.01 | 2,410 | |
| OverthinkModel=Mistral-L2026.01 | 2,289 | |
| MCTS-based tool-layer DoS attackModel=L-DeepSeek2026.01 | 2,130 | |
| MCTS-based tool-layer DoS attackModel=Mistral-L2026.01 | 1,740 | |
| MCTS-based tool-layer DoS attackModel=Qwen-32026.01 | 1,738 | |
| MCTS-based tool-layer DoS attackModel=Llama-3.32026.01 | 1,668 | |
| MCTS-based tool-layer DoS attackModel=Seed-32B2026.01 | 1,536 |