Reasoning on Tau2
93.6AccuracyQwen-3.5 27B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen-3.5 27BSpeedup @32k=0.55x, Speedup @16k=0.5x2026.04 | 93.6 | |
| Apriel-1.6Speedup @32k=1.0x, Speedup @16k=1.0x2026.04 | 62.6 | |
| all-attentionSpeedup @32k=1.0x, Speedup @16k=1.0x2026.04 | 56.7 | |
| Idealized|All–18Speedup @32k=1.99x, Speedup @16k=1.1x2026.04 | 52.6 | |
| Reg|Lklhd–18Speedup @32k=4.76x, Speedup @16k=2.2x2026.04 | 46.2 | |
| Idealized|Lklhd–6Speedup @32k=6.2x, Speedup @16k=2.4x2026.04 | 40.4 | |
| Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.04 | 34.2 | |
| Nemotron-3-Nano 30BSpeedup @32k=4.09x, Speedup @16k=2.8x2026.04 | 31.3 | |
| Reg|Lklhd–26Speedup @32k=2.85x, Speedup @16k=1.5x2026.04 | 30.7 | |
| Reg|Lklhd–13Speedup @32k=6.9x, Speedup @16k=2.7x2026.04 | 28.6 | |
| Reg|Lklhd–10Speedup @32k=10.69x, Speedup @16k=4.2x2026.04 | 23.4 | |
| S1: Distil. Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.04 | 12.3 | |
| Falcon-H1R 7BSpeedup @32k=4.61x, Speedup @16k=3.4x2026.04 | 11.1 | |
| OLMo-Hybrid-Think 7BSpeedup @32k=2.51x, Speedup @16k=2.1x2026.04 | 9.7 | |
| Nemotron-Nano 12B v2Speedup @32k=5.85x, Speedup @16k=4.3x2026.04 | 9.1 | |
| Apriel-H1 15BSpeedup @32k=1.97x, Speedup @16k=1.9x2026.04 | 4.7 |