Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Targeted retrieval on Needle-in-a-Haystack (NIAH)
Loading...
99.5
Mean Recall
Qwen3-14B
64.14
73.32
82.5
91.68
Jun 15, 2026
Mean Recall
Updated 1mo ago
Evaluation Results
Method
Method
Links
Mean Recall
Qwen3-14B
Training Source=Off-th...
2026.06
99.5
Qwen3-32B
Training Source=Off-th...
2026.06
99.3
CONTEXTRL
Training Source=Qwen3-...
2026.06
99
Base model
Training Source=Qwen3-...
2026.06
98.8
RL baseline
Training Source=Qwen3-...
2026.06
98.5
Qwen3-Coder-30B
Training Source=Off-th...
2026.06
85.7
CONTEXTRL
Training Source=Klear-...
2026.06
71.3
Base model
Training Source=Klear-...
2026.06
68.3
RL baseline
Training Source=Klear-...
2026.06
65.5
Feedback
Search any
task
Search any
task