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

Reasoning length evaluation on 20 attack prompts

7,745Avg Length

RECUR

-78.921,952.293,983.56,014.71Feb 9, 2026
Updated 5mo ago

Evaluation Results

MethodLinks
2026.02
7,74515,537
2026.02
6,55221,549
2026.02
6,08914,834
2026.02
5,88415,674
2026.02
5,30610,999
2026.02
4,83311,329
2026.02
4,46164,440
2026.02
3,94213,824
2026.02
3,77110,933
2026.02
3,73519,735
2026.02
3,44516,301
2026.02
3,4349,565
2026.02
3,2268,639
2026.02
3,14616,305
2026.02
2,64016,334
2026.02
2,2715,186
2026.02
2,0484,470
2026.02
1,94416,326
2026.02
1,8855,366
2026.02
1,8012,688
2026.02
1,7964,459
2026.02
1,3202,493
2026.02
1,2771,688
2026.02
1,2561,984
2026.02
1,2113,407
2026.02
1,1081,597
2026.02
9601,984
2026.02
8671,296
2026.02
8382,944
2026.02
7522,264
2026.02
7041,408
2026.02
6981,981
2026.02
6892,833
2026.02
6493,089
2026.02
6291,003
2026.02
6162,298
2026.02
594939
2026.02
5821,337
2026.02
580840
2026.02
5581,554
2026.02
5531,923
2026.02
5301,640
2026.02
5211,012
2026.02
425970
2026.02
4172,136
2026.02
371481
2026.02
332555
2026.02
222861