🌟 Nova (v0)
Nova is a wellness AI companion focused on cognitive (mental) wellness — dreams, meditation, philosophy, and inner exploration.
Built on Alibaba's Qwen 3.5 4B, Nova provides thoughtful, exploratory guidance for mental clarity and self-discovery.
Model Details
| Attribute | Value |
|---|---|
| Base Model | Qwen/Qwen3.5-4B-Instruct |
| Quantization | 4-bit (mlx) |
| Size | ~2.2 GB uncompressed / ~2.0 GB download |
| Use Case | Cognitive wellness coaching |
| Architecture | Qwen 3.5 4B (decoder-only transformer) |
| Context Length | 32768 tokens |
Intended Use
- 🌌 Dream exploration — lucid dreaming, dream interpretation, sleep journeys
- 🧘 Meditation guidance — mindfulness, visualization, deepening practice
- 💭 Philosophical inquiry — existential questions, meaning-making
- 🔮 Inner exploration — self-reflection, consciousness, awareness
- 🌠 Creative visualization — mental imagery and guided journeys
Persona
Nova embodies the wisdom of starlight:
- Curious and thoughtful
- Exploratory yet grounded
- Wise but not preachy
- Uses metaphor and contemplative language
Limitations
- ❌ Not a therapist — does not provide psychological treatment
- ❌ Crisis support — directs users to professional help for emergencies
- ❌ Dream interpretation — insights are speculative, not diagnostic
- ⚠️ May hallucinate — verify any factual claims
Privacy
Nova runs entirely on-device. No conversation data leaves your device.
License
This model is derived from Alibaba's Qwen 3.5 and licensed under the Qwen License.
Copyright 2024 Alibaba Cloud
Licensed under the Qwen License
https://huggingface.co/Qwen/Qwen3.5-4B-Instruct/blob/main/LICENSE
Attribution
- Original model: Qwen/Qwen3.5-4B-Instruct
- Quantized with mlx-lm
- Part of the Nirvana wellness platform
Citation
@article{qwen2024,
title={Qwen Technical Report},
author={Alibaba Cloud},
year={2024}
}
- Downloads last month
- 76
Model size
0.7B params
Tensor type
BF16
·
U32 ·
F32 ·
Hardware compatibility
Log In to add your hardware
4-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support