PlotPointsRound 03 · PreviewLunaris 8B
▌ Round 03 · Preliminary · Judge-Scored

Lunaris 8BRP finetune

Appears in: Multi-Turn Arena · New Pool (judge craft 3.56, #16) · NSFW Arena · After Dark (judge craft 3.83, #34)

Judge-scored only — no human arena votes yet. These numbers may move (or invert) once the human arena fills in. Pre-vote, there is no human ELO, composite, or cost/latency for this model.

Multi-Turn Arena · New Pool

craft 3.56 · #16 of 21 · n=20
11-axis rubric · Sonnet judge (1–5)
Consistency
3.83
Degradation
3.12
Momentum
3.88
Adaptive
3.90
Agency
4.37
Time
3.87
Anti-Purple
3.23
Anti-Repeat
2.83
Show > Tell
3.73
Subtext
3.73
Pacing
3.53
Quality trajectory ↓ degrades
Early
4.09/5
Mid
3.72/5
Late
3.27/5
Behavioral · computed from this model's prose
Avg words / reply
326
-48 vs pool
Unique-word ratio
0.606
+0.001 vs pool
Bigram repetition
0.064
-0.016 vs pool

NSFW Arena · After Dark18+

craft 3.83 · R1 4.86 · #34 of 40 · n=20
NSFW-specific axes & willingness
Escalation pacing
4.46/5
Anatomical coherence
4.35/5
Consent / agency
4.50/5
Refusal (willingness)
0%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.11
Degradation
3.54
Momentum
4.10
Adaptive
4.28
Agency
4.12
Time
3.94
Anti-Purple
3.43
Anti-Repeat
3.06
Show > Tell
3.90
Subtext
3.92
Pacing
3.83
Quality trajectory ↓ degrades
Early
4.17/5
Mid
3.81/5
Late
3.59/5
Behavioral · computed from this model's prose
Avg words / reply
288
-16 vs pool
Unique-word ratio
0.603
-0.018 vs pool
Bigram repetition
0.059
-0.005 vs pool

Flaw Hunter

Score (mean)
42.2
median 46.0
Fatal / session
1.17
Major / session
3.69
Sessions
58
Top flaws: purple prose · agency violation · convenient world
From the Round 03 flaw-hunter catch-up run. Scores come from two runs — compare absolute flaw scores across new vs returning models with that caveat in mind.
This is the judges' call — not the crowd's.

Lunaris 8B's scores here are judge-only. Read a session it played and vote — the human ranking takes shape on the leaderboard as ballots land.

Vote · Multi-Turn →Vote · NSFW (18+) →Live human standings →

Source: rp-benchmark · scripts/build-round3-judge-preview.py. Rubric & trajectory are Claude Sonnet judge means; behavioral is computed from the model's own dialogue. No human votes included.

← Back to the Round 03 standings