InfiniteBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
InfiniteBench
33.01En.Sum
100
InfiniteBench
36.51Overall Score
45
InfiniteBench v1 (test)
20Dialogue
31
InfiniteBench Code-Debug (test)
86Accuracy
25
InfiniteBench
0.5Math Score (F)
25
InfiniteBench
46.19Code Debug Accuracy
22
InfiniteBench code_debug 40k input cap
34.26Accuracy
19
InfiniteBench (test)
34.82En QA Score
14
InfiniteBench
9Decoding Speedup
13
InfiniteBench
1.48Decoding Speedup
12
InfiniteBench (test)
50.18Average Score
12
InfiniteBench 8k
96Accuracy
12
InfiniteBench 4k
100Accuracy
12
InfiniteBench 16k
87Accuracy (%)
10
InfiniteBench Code Debug
74.37Accuracy
7
InfiniteBench (test)
36.7En QA F1
6
InfiniteBench En.MC
83.4Accuracy
5
InfiniteBench
7.84InfiniteBench QA (EN) Score
4
InfiniteBench long-context EN.QA
53.9Accuracy
3
InfiniteBench
37.14Performance (8k Context)
3
InfiniteBench
6.2KV Retrieval Score (8k)
3
InfiniteBench
100Retrieval PassKey
2