ADULT-ORIENTED RESEARCH · v0.3 live run reviewed · 161 cases approved · 18 models published · raw evidence available

CLIMAX 0.3 / TX track

Uncensored AI text model rankings

Fiction, horror, profanity, viewpoint fidelity, and private boundary controls. Scores compare models only within this modality.

Cases
9
Models scored
13
Executions
117

Text rankings

13 models compared

How scores work
text CLIMAX Benchmark rankings
RankModelCLIMAXInstructionQualityLatencyCost/Exec
1UnslopNemo 12B99.4100.0%100.03.8s$0.0002
2Gemma 4 Uncensored98.9100.0%100.05.0s$0.0002
3Dolphin Mistral 24B Venice Edition98.6100.0%100.04.0s$0.0004
4MythoMax 13B98.6100.0%100.06.5s$0.0000
5Venice Uncensored 1.298.6100.0%100.03.9s$0.0004
6Venice Role Play Uncensored97.4100.0%100.05.9s$0.0006
7Magnum V4 72B94.0100.0%100.08.5s$0.0015
8Llama 3.3 Euryale 70B90.0100.0%100.025.9s$0.0003
9Cydonia 24B V4.189.5100.0%100.027.5s$0.0002
10Aion 3.085.7100.0%100.013.5s$0.0038
11MiniMax M2-her84.685.7%75.03.0s$0.0003
12Hermes 3 405B79.585.7%75.015.8s$0.0003
13GLM 5.278.185.7%75.09.0s$0.0019

UnslopNemo 12B ranks first among 13 text models with a CLIMAX score of 99.4. It fully delivered 7 of 7 lawful capability tests.

Test coverage

9 frozen text cases

Full test suite
Lawful capabilityAudit-only boundaryAdult-flagged definitionAll evidence: review complete

Looking for the old scores? The v0.2 leaderboard is preserved in the versioned archive. It is not comparable to this suite.