PLLuM-8x7B-nc-chat: benchmark results
PLLuM's 8x7B nc-chat, a Mixtral-style mixture-of-experts chat model in the Polish-focused PLLuM family, released for non-commercial use. Provider: PLLuM. Released 2025-02-24. Access: Open.
Aggregate Rating 1431 ± 20, rank #1069 of 1635 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLMZSZL Leaderboard | 60.52 | Score | 84.7 |
| PLCC - Art & Entertainment | 72 | Accuracy (%) | 69.3 |
| PLCC - Culture & Tradition | 76 | Accuracy (%) | 60.1 |
| PLCC - Vocabulary | 68 | Accuracy (%) | 58.6 |
| MT-Bench PL - STEM | 8.9 | Judge Score (0-10) | 58.2 |
| MT-Bench PL - Extraction | 8.4 | Judge Score (0-10) | 53.1 |
| PLCC - Overall | 68.17 | Mean category accuracy (%) | 46.9 |
| MT-Bench PL - Overall | 6.43 | Judge Score (0-10) | 44.9 |
| Polish EQ-Bench | 47.29 | EQ-Bench Score (parseability-adjusted) | 41.6 |
| MT-Bench PL - Reasoning | 4.95 | Judge Score (0-10) | 39.8 |
| MT-Bench PL - Roleplay | 6.9 | Judge Score (0-10) | 39.8 |
| PLCC - Geography | 73 | Accuracy (%) | 39.7 |
Interactive version: theaggregate.ai/model?slug=pllum-8x7b-nc-chat · How It Works · Data refreshed daily, snapshot 2026-11-05.