Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Runtime efficiency on NIAH (500 samples)
Loading...
34.01
Latency (8K Context)
H2O
33.6396
36.1398
38.64
41.1402
May 21, 2026
Latency (8K Context)
Latency (16K Context)
Latency (32K Context)
Latency (64K Context)
Latency (128K Context)
Latency (256K Context)
Updated 11d ago
Evaluation Results
Method
Method
Links
Latency (8K Context)
Latency (16K Context)
Latency (32K Context)
Latency (64K Context)
Latency (128K Context)
Latency (256K Context)
H2O
Input Length=2K tokens...
2026.05
34.01
34.72
37.15
40.34
44.46
53.37
SnapKV
Input Length=2K tokens...
2026.05
34.23
34.86
37.31
40.56
44.62
53.78
ZeroMerge
Input Length=2K tokens...
2026.05
34.87
35.53
38.07
41.21
45.35
54.62
Judge Q
Input Length=2K tokens...
2026.05
35.12
35.86
38.43
41.72
45.94
55.29
Meta-Soft
Input Length=2K tokens...
2026.05
35.29
36.03
38.72
42.05
46.36
55.78
Full KV
Input Length=2K tokens...
2026.05
43.27
54.38
79.93
142.95
271.83
583.41
Feedback
Search any
task
Search any
task
