Stateful Service Dialogue on tau-bench Airline
58Task Completion Score (TCS)Gemini-3-flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-flashAverage Steps (Stp)=9.0, Input Tokens (Tok In)=77.6, Output Tokens (Tok Out)=6.8, Execution Time=233.02026.05 | 58 | |
| Qwen3-235B-A22BAverage Steps (Stp)=11.2, Input Tokens (Tok In)=86.0, Output Tokens (Tok Out)=1.4, Execution Time=54.02026.05 | 52 | |
| GPT-5-miniAverage Steps (Stp)=16.0, Input Tokens (Tok In)=112.2, Output Tokens (Tok Out)=32.1, Execution Time=1022.72026.05 | 46 | |
| DeepSeek-V3.2Average Steps (Stp)=18.8, Input Tokens (Tok In)=128.3, Output Tokens (Tok Out)=3.1, Execution Time=236.92026.05 | 38 |