ghosty — on-device models (Core ML, iOS 18+)
Models for the ghosty iPhone app by Yehor Lisovets, fine-tuned from Qwen3 (Apache 2.0) and converted to Core ML (stateful KV cache, 4-bit weights). ghosty81 ← Qwen3-1.7B, ghosty81pro ← Qwen3-4B, ghosty81max ← Qwen3-8B. Older ghosty8 / ghosty8pro files are kept for older app builds. Modified by Yehor Lisovets (fine-tuning and conversion).
- Downloads last month
- 19
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support