PLLuM-12B-nc-chat-250715: benchmark results
PLLuM's 12B nc-chat (2507) build, a chat model in the Polish-focused PLLuM family, released for non-commercial use. Provider: PLLuM. Released 2025-07-23. Access: Open.
Aggregate Rating 1454 ± 17, rank #953 of 1635 rated models, from 64 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLMZSZL Leaderboard | 54.24 | Score | 70.4 |
| PLCC - Art & Entertainment | 72 | Accuracy (%) | 69.3 |
| CPTU Bench | 3.67 | Average Score (1-5) | 63.4 |
| Polish EQ-Bench | 55.17 | EQ-Bench Score (parseability-adjusted) | 58.4 |
| PLCC - Culture & Tradition | 75 | Accuracy (%) | 57.6 |
| PLCC - Vocabulary | 67 | Accuracy (%) | 56.6 |
| PLCC - Overall | 69.67 | Mean category accuracy (%) | 49.6 |
| PLCC - Geography | 79 | Accuracy (%) | 48.8 |
| Polish Knowledge Benchmark - Music | 70.35 | Accuracy (%) | 47.1 |
| Belebele European - German | 69.78 | Accuracy (%, 0-shot) | 46.4 |
| Polish Knowledge Benchmark - Film & Theatre | 70.82 | Accuracy (%) | 45.6 |
| Polish Knowledge Benchmark - Theatre | 67.69 | Accuracy (%) | 44.9 |
Interactive version: theaggregate.ai/model?slug=pllum-12b-nc-chat-250715 · How It Works · Data refreshed daily, snapshot 2026-11-05.