The new sovereign brain (rank=32 LoRA, trained in Phase 29) is wired and live.
GET /api/sovereign-brain/v2/status
{
"brain": "SovereignBrainV2",
"base_model": "Qwen3-0.6B",
"rank": 32,
"lora_alpha": 64,
"target_modules": ["q_proj", "k_proj", "v_proj", "o_proj"],
"adapter_exists": true,
"adapter_size_mb": ~38,
"phase": 29,
"loss_reduction_pct": 57.7
}
POST /api/sovereign-brain/v2 with {"message": "..."} โ returns SIGIL-signed sovereign answer in ~7s
| OWEM | Specialty | Loss Reduction | Rank |
|---|---|---|---|
| compliance | Article 0 + 12 Pillars + BFT-33 + SIGIL | 83.5% | 16 |
| defense | DEFONEOS + AUKUS + hard stops | 86.2% | 16 |
| intuition | World model + JEPA + emergence | 78.6% | 16 |
| voice | Voice + privacy + 12 Pillars | 89.3% | 16 |
| general (sov_brain_v2) | General sovereign fallback | 57.7% | 32 |
Avg time: 11.0s ยท Success: 10/10 ยท Rank=32 brain behind all OWEMs
[COMPLIANCE] โ What is Article 0? A: Article 0 = legal binding agreement. Includes: sovereign, no BFT-33... Time: 20.2s ยท SIGIL: f50db74ec9f4dfdb [DEFENSE] โ What are the 3 DEFONEOS compartments? A: DEFONEOS compartments = defense OWEM's compartments... [INTUITION] โ How does the world model detect OOD? A: Sovereign world model detects OOD using OOD detection metric... [VOICE] โ How does SOV33 handle voice privacy? A: SOV33 uses 12 Pillars of Sovereignty. Each response is encrypted... [GENERAL] โ What is the sovereign substrate? A: Sovereign substrate = the foundation for sovereign AI...