//opengauntlet
← back to the leaderboard

Gemma 4 31B Dark Thoughts V2 (Q4_K_M, no thinking) conversational AI benchmark

gemma4-31b-darkthoughts-v2-q4-nothink

Gemma 4 31B Dark Thoughts V2 (Q4_K_M, no thinking) benchmark: humanlikeness scores, pairwise rankings, transcripts, latency, and measured local speed.

Leaderboard rank
#24 of 41
Humanlikeness
74.3 / 100
Pairwise rating
1556

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

Loading model…