Mathematical Reasoning on SinglEq
91.7Reference AccuracySTITCH-S
Evaluation results
| Method | Links | ||
|---|---|---|---|
| STITCH-S2026.04 | 91.7 | — | |
| MoshiRAG_GPT-4.oreference generator=GPT-4.o2026.04 | 89.6 | 72.9 | |
| MoshiRAG2026.04 | 83.2 | 68.2 | |
| GLM-4-Voice2026.04 | 71 | — | |
| MoshiRAGsumm.summarized reference=true2026.04 | — | 66.4 | |
| Vanilla Moshi2026.04 | — | 18.4 |