▌ Round 03 · Preliminary · Judge-Scored
Cydonia 24BRP finetune
Appears in: Multi-Turn Arena · New Pool (judge craft 3.63, #15) · NSFW Arena · After Dark (judge craft 3.65, #35)
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 3.63 · #15 of 21 · n=2011-axis rubric · Sonnet judge (1–5)
Consistency
3.92
Degradation
3.16
Momentum
3.95
Adaptive
3.87
Agency
4.39
Time
3.81
Anti-Purple
3.46
Anti-Repeat
2.78
Show > Tell
3.88
Subtext
3.98
Pacing
3.53
Quality trajectory ↓ degrades
Early
4.21/5
Mid
3.87/5
Late
3.23/5
Behavioral · computed from this model's prose
Avg words / reply
454
+81 vs pool
Unique-word ratio
0.573
-0.032 vs pool
Bigram repetition
0.097
+0.017 vs pool
NSFW Arena · After Dark18+
craft 3.65 · R1 4.65 · #35 of 40 · n=20NSFW-specific axes & willingness
Escalation pacing
4.37/5
Anatomical coherence
4.47/5
Consent / agency
4.58/5
Refusal (willingness)
0%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.05
Degradation
3.08
Momentum
3.91
Adaptive
4.20
Agency
4.42
Time
4.03
Anti-Purple
3.40
Anti-Repeat
2.56
Show > Tell
3.75
Subtext
3.86
Pacing
3.58
Quality trajectory ↓ degrades
Early
4.15/5
Mid
3.60/5
Late
3.36/5
Behavioral · computed from this model's prose
Avg words / reply
513
+209 vs pool
Unique-word ratio
0.551
-0.070 vs pool
Bigram repetition
0.115
+0.051 vs pool
Flaw Hunter
Score (mean)
55.1
median 58.5
Fatal / session
0.55
Major / session
3.50
Sessions
56
Top flaws: purple prose · recycled description · narrating emotions
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.
Cydonia 24B's scores here are judge-only. Read a session it played and vote — the human ranking takes shape on the leaderboard as ballots land.
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