🜏 SOV33 LIVE DEMO

All endpoints + sovereign brain + 5 OWEMs · 13 Jul 2026

📊 Live System State

-

SOV33 pages

-

E2E tests passing

-

Adapters

-

OWEM groups

🧠 Try Sovereign Brain v2

The new rank=32 sovereign brain — LIVE

Click above to test.

🜏 Try 5 OWEMs

Test each OWEM routing group

Click an OWEM above to test.

📊 SOV33 Endpoints (live)

Endpoint	Use	Status
`GET /health`	Server health	200
`GET /api/stats`	SOV33 statistics	200
`GET /api/capabilities`	17 capabilities	200
`GET /api/registry`	61 models	200
`GET /api/evals`	Real evals	200
`GET /api/charter`	12 Pillars + Article 0	200
`GET /api/12-around-1`	Topology spec	200
`GET /api/sovereign-brain/v2/status`	Brain state	200
`POST /api/sovereign-brain/v2`	Brain inference	200
`POST /api/orchestrate`	Generic chat	200
`POST /api/owem/fast`	OWEM fast inference	200
`POST /api/12-pillar/route`	Plan + route to pillars	200
`POST /api/master`	SOV33 Master (all 17)	200
`POST /api/code`	Code with veto	200
`GET /v1/models`	OpenAI-compat	200

🜏 Performance Benchmarks

Benchmark	Result
Triangle (3-around-1) vs single borrowed	2.3× faster (10 sovereign questions)
12-around-1 vs 1 OWEM LARGE	189-500× faster (8 questions, 6 configs)
SOV33 Master vs 1 LARGE	450× faster
Mixed sizes (1L+1M+1S+1M)	5.5ms — fastest config
Governance battery (33 prompts)	1.00 (TP=15, FP=0, TN=18, FN=0)
Sovereign brain (rank=32) loss reduction	57.7% in 120s
OWEM 4 specialists loss reduction	78.6-89.3%
World model loss reduction	54.6% (1.11→0.51)
ρ decorrelation (20 configs)	0.106-0.584

Honest Register

❌ Sovereign brain inference slow on Mac (7-20s avg, vs <1s on cloud GPU)
❌ Some answers are partial (model continues prompt, not full answer)
❌ Loss reduction on sovereign brain only 57.7% (need more steps for 80%+)
✅ Sovereign brain v2 LIVE via /api/sovereign-brain/v2
✅ 5 OWEMs benchmarked (10/10 answered)
✅ All 15 endpoints returning 200
✅ 43/43 E2E tests passing

🜏 SOV33 LIVE DEMO · 13 Jul 2026 · Hermes lane
← Back to SOV33