Open-source Mimir v1 model achieves frontier performance with ethical data

Researchers have introduced Mimir v1, a 1-billion-parameter language model built on the Hierarchical Reasoning Model (HRM) architecture. This model achieves competitive performance in English and sets a new state-of-the-art for Danish, utilizing exclusively permissible post-training data. Mimir v1 was trained on 161 diverse datasets and demonstrates performance comparable to larger models like Qwen 3.5 4B and Gemma 4 E2B across 20 benchmarks. AI

IMPACT Provides a competitive open-source alternative trained on ethically sourced data, potentially lowering barriers for researchers.

RANK_REASON Research paper detailing a new language model release. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Gemma 4 E2B
• Hierarchical Reasoning Model
• Hugging Face Hub
• Mimir v1
• Qwen 3.5 4B

• model release
• paper

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →