benchgap
Calibration

HLE-MultiChoice (Decision Index) → Artificial Analysis Decision Index v1.0

Artificial Analysis Decision Index v1.0 is estimated from HLE-MultiChoice (Decision Index) with a Michaelis–Menten curve fitted on 19 models measured on both: y = 1.0109·x / (0.10653 + x), R² = 0.68, cross-validated error 4.2 pp. It is used for 0 estimates.

Fitted on

ModelHLE-MultiChoice (Decision Index)Artificial Analysis Decision Index v1.0
Celeris-1.1 Decision12.5%60.9%
Celeris-1 Decision13.3%52.3%
Clef11.1%48.6%
Clef-Flash10.3%45.8%
Clef-Omni9.0%44.3%
d1-ultra15.0%65.2%
Drex Max16.0%63.7%
Drex Max v1.117.0%63.8%
Drex v1.512.1%50.7%
Drex v2.013.6%57.9%
elephant-118.3%64.7%
GLiDE10.5%53.9%
GLiDE v1.217.5%65.0%
Jev 1.1418.9%57.1%
d111.5%50.3%
Mercury Decide10.1%52.7%
Microsoft-Decision-110.9%53.8%
Perplexity Decider Flash v1.2 4B12.1%50.2%
Perplexity Decider v1.2 27B14.6%56.2%