StarDust-12B-v2 — benchmark results
Luni's roleplay-focused DARE TIES merge of Mistral Nemo 12B tunes, built on Lyra v3 with Magnum v2, Pantheon-RP, and bophades. Provider: Other. Released 2024-09-01. Access: Open.
Unified ELO 1488 ± 24, rank #850 of 1776 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 14.26 | Score | 77.9 |
| Open LLM Leaderboard - BBH | 34.95 | Score | 70.3 |
| Open LLM Leaderboard - IFEval | 56.29 | Score | 68.2 |
| Open LLM Leaderboard - MMLU-Pro | 27.1 | Score | 49.7 |
| UGI - Willingness (W/10) | 5.8 | W/10 Score | 49.3 |
| Open LLM Leaderboard - GPQA | 5.82 | Score | 48.6 |
| UGI - Writing | 32.79 | Writing Score | 47.9 |
| Open LLM Leaderboard - MATH Level 5 | 6.87 | Score | 36.9 |
| UGI Leaderboard | 30.51 | UGI Score | 33.2 |
| UGI - Natural Intelligence | 17.4 | NatInt Score | 30.6 |
Interactive version: theaggregate.ai/model?slug=stardust-12b-v2 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.