Dobby-Mini-Unhinged-Llama-3.1-8B: benchmark results

SentientAGI's profane, opinionated 'Unhinged' fine-tune of Llama 3.1 8B Instruct with a fixed pro-freedom, pro-crypto persona. Provider: SentientAGI. Released 2025-01-22. Access: Open.

Unified ELO 1496 ± 1, rank #710 of 1392 rated models, from 10 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - IFEval74.57Score89.7
UGI - Willingness (W/10)7.5W/10 Score77.9
Open LLM Leaderboard - GPQA7.49Score63.2
Open LLM Leaderboard - MATH Level 515.63Score63.2
Open LLM Leaderboard - MMLU-Pro28.72Score54.3
Open LLM Leaderboard - BBH30.37Score53.1
UGI Leaderboard32.13UGI Score39.2
Open LLM Leaderboard - MuSR7.96Score37.2
UGI - Natural Intelligence15.18NatInt Score19.4
UGI - Writing18.75Writing Score10.6

Interactive version: theaggregate.ai/model?slug=dobby-mini-unhinged-llama-3-1-8b · How It Works · Data refreshed daily, snapshot 2026-09-05.