Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Prefill Throughput on 96K context
Loading...
7,959.4
Prefill Throughput (tok/s)
SparDA
4,887.032
5,684.666
6,482.3
7,279.934
Jun 3, 2026
Prefill Throughput (tok/s)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Prefill Throughput (tok/s)
SparDA
Backbone=MiniCPM4.1-8B...
2026.06
7,959.4
Sparse
Backbone=MiniCPM4.1-8B...
2026.06
6,776.8
Sparse
Backbone=MiniCPM4.1-8B...
2026.06
6,765.9
InfiniGen
Backbone=MiniCPM4.1-8B...
2026.06
6,740
SparDA
Backbone=NOSA-8B, Hard...
2026.06
5,651.5
Dense
Backbone=NOSA-8B, Hard...
2026.06
5,313.5
Dense
Backbone=MiniCPM4.1-8B...
2026.06
5,260.2
Sparse
Backbone=NOSA-8B, Hard...
2026.06
5,022.4
Sparse
Backbone=NOSA-8B, Hard...
2026.06
5,021.9
InfiniGen
Backbone=NOSA-8B, Hard...
2026.06
5,005.2
Feedback
Search any
task
Search any
task
