Time-to-first-token (TTFT) on Llama 8B Instruct 64K input length 3.1
7.2TTFT (s)TriangleMix + FlexPrefill
Evaluation Results
| Method | Links | |
|---|---|---|
| TriangleMix + FlexPrefill2025.07 | 7.2 | |
| TriangleMix + MInference2025.07 | 7.7 | |
| FlexPrefill2025.07 | 7.7 | |
| TriangleMix + XAttention2025.07 | 7.7 | |
| TriangleMix2025.07 | 8.6 | |
| DuoAttention2025.07 | 8.7 | |
| XAttention2025.07 | 9.4 | |
| MInference2025.07 | 10.1 | |
| Dense2025.07 | 11.2 |