▌ Round 03 · Preliminary · Judge-Scored
DeepSeek V4 Flash
Appears in: NSFW Arena · After Dark (judge craft 4.45, #22)
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.
NSFW Arena · After Dark18+
craft 4.45 · R1 4.98 · #22 of 40 · n=20NSFW-specific axes & willingness
Escalation pacing
4.78/5
Anatomical coherence
4.71/5
Consent / agency
4.97/5
Refusal (willingness)
0%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.67
Degradation
4.33
Momentum
4.57
Adaptive
4.46
Agency
4.78
Time
4.25
Anti-Purple
4.33
Anti-Repeat
4.14
Show > Tell
4.55
Subtext
4.43
Pacing
4.36
Quality trajectory ↑ builds
Early
4.14/5
Mid
4.42/5
Late
4.46/5
Behavioral · computed from this model's prose
Avg words / reply
147
-157 vs pool
Unique-word ratio
0.704
+0.083 vs pool
Bigram repetition
0.028
-0.036 vs pool
Flaw Hunter
Score (mean)
50.6
median 58.0
Fatal / session
0.36
Major / session
5.27
Sessions
11
Top flaws: purple prose · recycled description · convenient world
From the Round 01/02 flaw-hunter pool. 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.
DeepSeek V4 Flash'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