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

Needle-in-the-haystack

Benchmarks

Task NameDataset NameSOTA ResultTrend
Long-context retrievalNeedle-in-the-Haystack 10k-context
Accuracy100
30
Needle-in-the-haystack RetrievalNeedle-in-the-haystack 2500 tokens
Needle Accuracy100
16
Showing 2 of 2 rows