Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Time-to-first-token (TTFT) on Llama 8B Instruct 48K input length 3.1
Loading...
5.2
TTFT (s)
TriangleMix + FlexPrefill
5.096
5.798
6.5
7.202
Jul 29, 2025
TTFT (s)
Updated 1mo ago
Evaluation Results
Method
Method
Links
TTFT (s)
TriangleMix + FlexPrefill
2025.07
5.2
FlexPrefill
2025.07
5.5
TriangleMix + XAttention
2025.07
5.5
DuoAttention
2025.07
5.8
TriangleMix
2025.07
5.9
TriangleMix + MInference
2025.07
6
XAttention
2025.07
6.6
Dense
2025.07
7.3
MInference
2025.07
7.8
Feedback
Search any
task
Search any
task