Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Needle-in-a-Haystack

Benchmarks

Task NameDataset NameSOTA ResultTrend
Long-context retrievalNeedle-in-a-Haystack (test)
Accuracy100
56
Needle-in-a-HaystackNeedle-in-a-Haystack
Accuracy100
44
Needle RetrievalNeedle In A Haystack
Exact-match Accuracy67
40
Needle-in-a-haystackNeedle-in-a-haystack 4x original context
Accuracy100
35
Needle-in-a-Haystack RetrievalNeedle-in-a-Haystack 32K context (test)
Accuracy76
30
Needle-in-a-Haystack RetrievalNeedle-in-a-Haystack 8K context (test)
Accuracy100
30
Long-context RetrievalNeedle-in-a-Haystack
Retrieval Accuracy100
29
RetrievalNeedle-in-a-Haystack L=8k
Accuracy100
24
RetrievalNeedle-in-A-Haystack RULER
Success Rate (4K Context)100
22
Key Information RetrievalNeedle-in-a-Haystack 32K context
Accuracy98.2
19
RetrievalNeedle-in-a-Haystack Induction
Accuracy98
18
Multi-key needle-in-a-haystack recallMulti-key needle-in-a-haystack 16k context length
Recall100
16
Long-context RetrievalNeedle-in-a-Haystack 64K-token context
Success Rate98.7
12
Targeted retrievalNeedle-in-a-Haystack (NIAH)
Mean Recall99.5
9
Synthetic RetrievalNeedle-In-A-Haystack (NIAH)
NIAH-1 Success Rate99.8
7
Long-context Information RetrievalNeedle-In-a-Haystack Verbatim prompt (test)
Accuracy (Depth 0%)0.996
7
Needle-In-a-HaystackNeedle-In-a-Haystack Gemini prompt (test)
Success Rate @ 0% Insertion57.2
7
Information RetrievalNeedle In A Haystack
Recall@1K90
6
Exact RetrievalNeedle-in-a-Haystack (NIAH) 64K
Average Accuracy97.6
5
Exact RetrievalNeedle-in-a-Haystack (NIAH) 32K
Average Accuracy100
5
Exact RetrievalNeedle-in-a-Haystack (NIAH) 16K
Average Accuracy100
5
Long-context retrievalNeedle-in-a-Haystack 1.0 (test)
Score99.9
5
Needle-in-a-haystackNeedle-in-a-haystack 8x original context
Accuracy52.2
4
Needle-in-a-haystackNeedle-in-a-haystack 2x original context
Needle-in-a-haystack Accuracy (2x Context)74.92
4
Information RetrievalNeedle-in-a-Haystack
Average Score98.2
3
Showing 25 of 27 rows