//opengauntlet
← back to the leaderboard

Versipellis 31B (Q4_K_M, no thinking) conversational AI benchmark

versipellis-31b-q4-nothink

Versipellis 31B (Q4_K_M, no thinking) benchmark: humanlikeness scores, pairwise rankings, transcripts, latency, and measured local speed.

Leaderboard rank
#18 of 41
Humanlikeness
76.3 / 100
Pairwise rating
1615

Evidence: identical scenario pack and judge protocol; measured and judged results are shown separately below. Read the methodology.

Loading model…