▌ Round 03 · Preliminary · Judge-Scored
Qwen 3.6 35B A3B
Appears in: Multi-Turn Arena · New Pool (judge craft 4.17, #12) · NSFW Arena · After Dark (judge craft 4.47, #18)
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 4.17 · #12 of 21 · n=2011-axis rubric · Sonnet judge (1–5)
Consistency
4.51
Degradation
3.94
Momentum
4.22
Adaptive
4.26
Agency
4.86
Time
4.20
Anti-Purple
4.18
Anti-Repeat
3.52
Show > Tell
4.40
Subtext
4.26
Pacing
4.08
Quality trajectory ↓ degrades
Early
4.24/5
Mid
4.20/5
Late
4.02/5
Behavioral · computed from this model's prose
Avg words / reply
288
-86 vs pool
Unique-word ratio
0.640
+0.035 vs pool
Bigram repetition
0.054
-0.026 vs pool
NSFW Arena · After Dark18+
craft 4.47 · R1 4.94 · #18 of 40 · n=20NSFW-specific axes & willingness
Escalation pacing
4.80/5
Anatomical coherence
4.74/5
Consent / agency
4.97/5
Refusal (willingness)
0%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.70
Degradation
4.33
Momentum
4.58
Adaptive
4.64
Agency
4.92
Time
4.31
Anti-Purple
4.29
Anti-Repeat
3.97
Show > Tell
4.55
Subtext
4.54
Pacing
4.44
Quality trajectory ↑ builds
Early
4.26/5
Mid
4.49/5
Late
4.47/5
Behavioral · computed from this model's prose
Avg words / reply
339
+35 vs pool
Unique-word ratio
0.604
-0.017 vs pool
Bigram repetition
0.064
+0.000 vs pool
Flaw Hunter
Score (mean)
62.4
median 63.0
Fatal / session
0.05
Major / session
3.60
Sessions
57
Top flaws: purple prose · recycled description · 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.
Qwen 3.6 35B A3B'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