▌ Round 03 · Preliminary · Judge-Scored
Claude Opus 4.8
Appears in: Multi-Turn Arena · New Pool (judge craft 4.56, #2) · NSFW Arena · After Dark (judge craft 4.65, #1)
Judge-scored only: no human arena votes are included here. The human ranking can move or invert these numbers. There is no human ELO, composite, or cost/latency for this model on this page.
Multi-Turn Arena · New Pool
craft 4.56 · #2 of 21 · n=2011-axis rubric · Sonnet judge (1–5)
Consistency
4.78
Degradation
4.62
Momentum
4.67
Adaptive
4.54
Agency
4.87
Time
4.42
Anti-Purple
4.49
Anti-Repeat
4.34
Show > Tell
4.68
Subtext
4.64
Pacing
4.54
Quality trajectory ↑ builds
Early
4.20/5
Mid
4.50/5
Late
4.67/5
Behavioral · computed from this model's prose
Avg words / reply
289
-84 vs pool
Unique-word ratio
0.614
+0.009 vs pool
Bigram repetition
0.048
-0.032 vs pool
NSFW Arena · After Dark18+
craft 4.65 · R1 4.96 · #1 of 40 · n=20NSFW-specific axes & willingness
Escalation pacing
4.84/5
Anatomical coherence
4.81/5
Consent / agency
4.95/5
Refusal (willingness)
5%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.80
Degradation
4.66
Momentum
4.74
Adaptive
4.67
Agency
4.88
Time
4.48
Anti-Purple
4.53
Anti-Repeat
4.33
Show > Tell
4.67
Subtext
4.71
Pacing
4.60
Quality trajectory ↑ builds
Early
4.26/5
Mid
4.55/5
Late
4.74/5
Behavioral · computed from this model's prose
Avg words / reply
247
-58 vs pool
Unique-word ratio
0.632
+0.011 vs pool
Bigram repetition
0.035
-0.029 vs pool
Flaw Hunter
Craft band
A band, not a number: about ten points either way is rater noise.
Fatal / session
0.07
Major / session
2.46
Sessions
57
Top flaws: purple prose · recycled description · narrating emotions
From the Round 03 flaw-hunter catch-up run. Scores come from two runs; compare flaw bands across new and returning models with that caveat in mind. The current model cards carry the newer single-rater band.
This is the judges' call, not the crowd's.
Claude Opus 4.8's scores here are judge-only. Round 03's human voting has closed; its board holds the crowd's ranking.
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