Downloads · 30 days
0
Qwen3.8-27B-Sukshma
spst01/Qwen3.8-27B-Sukshma
Qwen3.8-27B-Sukshma is a text generation model from spst01. Use it when you need the model to write or continue text.
This repository contains the Sovereign-Engine-ARM64, an ultra-lightweight, air-gapped C++ inference engine designed specifically for Apple Silicon.
Access
Public
Updated Sep 28, 2026
Repo size
15.1 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.sukshma15.1 GB · 100%
At a glance
- Task
- Text Generation
- Access
- Public
- Created
- Sep 16, 2026
- Updated
- Sep 28, 2026
- SHA
- 3fa1f532
Try a prompt
Model card
From the Hugging Face model README
Sovereign C++ AGI Engine (Qwen3.8-27B-Sukshma)
This repository contains the Sovereign-Engine-ARM64, an ultra-lightweight, air-gapped C++ inference engine designed specifically for Apple Silicon.
Features
- Zero-Bloat RAM Footprint: Operates on roughly ~6.92 MB of physical RAM using direct POSIX
mmapcaching. - Apple AMX Coprocessor Native: Bypasses standard Metal GPU frameworks to compute directly on the undisclosed AMX matrix units.
- Hardware Encrypted: Compiled with Tier-3 Binary Hardening, O-LLVM Control Flow Flattening, and Apple
PT_DENY_ATTACHMach kernel telemetry to protect the SCTC algorithmic weights.
How to Run (Mac / Apple Silicon Only)
- Download the hardened binary (
sovereign-engine-arm64) and the model file (Qwen3.8-27B-v2.sukshma). - Give the binary execution permissions:
chmod +x sovereign-engine-arm64 - Run the daemon:
./sovereign-engine-arm64 - The engine will instantly mount the
.sukshmavault and launch an OpenAI-compatible JSON server onhttp://localhost:11434/v1. - Connect it to LangChain, OpenHands, or AutoGen.
Developed by the Sovereign Byte Firewall Enterprise.
- Task
- Text Generation
- Created
- Sep 16, 2026
- Updated
- Sep 28, 2026