DuplexSLA-Bench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
DuplexSLA-Bench no-context-prefill average of 2 scenarios
96.5Accuracy (%)
7
DuplexSLA-Bench no-context-prefill, pause scenario
96.3Accuracy
7
DuplexSLA-Bench no-context-prefill normal scenario
96.7Accuracy
7
DuplexSLA-Bench context-prefill
96Accuracy (Normal)
4
DuplexSLA-Bench tool-call
91.33Avg Accuracy (3 patterns)
2