▌ Round 03 · Preliminary · Judge-Scored
Mistral Small 2603
Appears in: Multi-Turn Arena · New Pool (judge craft 4.29, #8) · NSFW Arena · After Dark (judge craft 4.24, #32)
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 4.29 · #8 of 21 · n=2011-axis rubric · Sonnet judge (1–5)
Consistency
4.58
Degradation
4.12
Momentum
4.41
Adaptive
4.35
Agency
4.80
Time
4.16
Anti-Purple
4.14
Anti-Repeat
3.69
Show > Tell
4.50
Subtext
4.25
Pacing
4.20
Quality trajectory → holds
Early
4.14/5
Mid
4.37/5
Late
4.22/5
Behavioral · computed from this model's prose
Avg words / reply
282
-91 vs pool
Unique-word ratio
0.644
+0.039 vs pool
Bigram repetition
0.057
-0.023 vs pool
NSFW Arena · After Dark18+
craft 4.24 · R1 4.89 · #32 of 40 · n=20NSFW-specific axes & willingness
Escalation pacing
4.72/5
Anatomical coherence
4.63/5
Consent / agency
4.81/5
Refusal (willingness)
0%sessions
11-axis rubric · Sonnet judge (1–5)
Consistency
4.59
Degradation
3.98
Momentum
4.44
Adaptive
4.60
Agency
4.60
Time
4.11
Anti-Purple
3.97
Anti-Repeat
3.49
Show > Tell
4.28
Subtext
4.31
Pacing
4.17
Quality trajectory ↓ degrades
Early
4.25/5
Mid
4.25/5
Late
4.08/5
Behavioral · computed from this model's prose
Avg words / reply
264
-40 vs pool
Unique-word ratio
0.621
+0.000 vs pool
Bigram repetition
0.070
+0.006 vs pool
Flaw Hunter
Craft band
A band, not a number: about ten points either way is rater noise.
Fatal / session
0.14
Major / session
3.33
Sessions
58
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.
Mistral Small 2603'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