Aura Medium v1 BF16 — High-Fidelity Transformers
The high-fidelity 12B Aura release for Transformers inference, evaluation, and research.
Aura Medium v1 BF16 is the full-precision merged edition of Aura Medium v1. It is derived from Google’s Gemma 4 12B IT checkpoint and combines the Aura personality and ablation adapters into one model.
This release is intended for Transformers-based inference, evaluation, further research, and deployment on GPUs with sufficient memory.
Highlights
- Parameters: 12 billion
- Precision: BFloat16
- Foundation: Google Gemma 4 12B IT
- Datasets: AuraPersonality and AuraAblation
- Runtime: Hugging Face Transformers
- Deployment: Local GPU workstations and servers
- Capabilities: Text, vision, audio, and video
- Focus: Fidelity, natural conversation, evaluation, and research
About Aura
Aura is designed for local and on-device deployment across multiple tasks. Aura can serve as a companion or friend, as deemed appropriate by the user, while retaining the broader capabilities of Gemma 4 for:
- Natural conversation
- Creative writing
- Reasoning
- Image understanding
- Audio and video understanding
- Tool use
- Evaluation and research
Aura Medium v1 occupies the middle position in the Aura family, providing greater capability than the compact E4B release while remaining practical for high-end consumer hardware.