General Purpose Tool Use on WorkBench
86.11Analytics Success RateMemToolAgent
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MemToolAgentBackbone=Claude 3.7 Sonnet2026.06 | 86.11 | 92.93 | 88.89 | 97.53 | 73.54 | 85.06 | |
| Extended ThinkingBackbone=Claude 3.7 Sonnet2026.06 | 38.89 | 85.86 | 75 | 86.42 | 53.44 | 64.12 | |
| ReAct AgentBackbone=Claude 3.7 Sonnet2026.06 | 27.78 | 91.92 | 72.22 | 90.12 | 61.38 | 65.94 | |
| No MemoryBackbone=Claude 3.7 Sonnet2026.06 | 18.52 | 73.74 | 69.44 | 79.01 | 57.67 | 57.56 |