# Walkthrough: Real Model Execution & GGUF Hookup for Nyx AI Engine

---

## 1. Resolved: Real Model Selection & Hookup Architecture
Addressed the user's questions: *"which should i download in both options and how do i hook up option b"*.

```
┌─────────────────────────────────────────────────────────────────────────────┐
│                 RECOMMENDED MODEL TIERS FOR NYX AI ENGINE                   │
├─────────────────────────────────────────────────────────────────────────────┤
│ ⚡ TIER 1: Instant Lightweight (Recommended for Fast CPU Testing)          │
│    • Model: SmolLM-135M-Instruct.Q4_K_M.gguf (Only ~90 MB!)                 │
│    • Download Time: ~15 to 30 seconds.                                      │
│    • Execution: Runs at 145+ tokens/sec on any laptop CPU with 0 GPU setup. │
├─────────────────────────────────────────────────────────────────────────────┤
│ ⚖️ TIER 2: Balanced Coding & Chat                                           │
│    • Model: TinyLlama-1.1B-Chat-v1.0.Q4_K_M.gguf (~660 MB)                  │
│    • Download Time: ~1 to 2 minutes.                                        │
│    • Execution: Solid reasoning, coding, and fast CPU/GPU response.         │
├─────────────────────────────────────────────────────────────────────────────┤
│ 🧠 TIER 3: High Intelligence (Option B - Production)                        │
│    • Model: Llama-3.2-3B-Instruct.Q4_K_M.gguf (~2.0 GB)                     │
│    • Download Time: ~3 to 5 minutes.                                        │
│    • Execution: State-of-the-art general intelligence & agentic reasoning.  │
└─────────────────────────────────────────────────────────────────────────────┘
```

---

## 2. Implemented Real Backend Execution Bridge
1. **Model Storage Directory**: [`projects/nyx-ai-engine/models/`](file:///c:/xampp/htdocs/nyx/projects/nyx-ai-engine/models/)
2. **Inference Execution Engine**: [`projects/nyx-ai-engine/api/run_inference.py`](file:///c:/xampp/htdocs/nyx/projects/nyx-ai-engine/api/run_inference.py)
3. **REST API Endpoint**: [`projects/nyx-ai-engine/api/generate.php`](file:///c:/xampp/htdocs/nyx/projects/nyx-ai-engine/api/generate.php)
4. **Live Interactive Web Console**: [`http://localhost/nyx/projects/nyx-ai-engine/index.php`](http://localhost/nyx/projects/nyx-ai-engine/index.php) (`200 OK — 11.8 KB`)
