A
APEX v3.2
HUB-SATELLITE ยท 11โ†’42 AGENTS ยท COPPA-GRADE
11 LAUNCHHUB+SATFREE VOICENO HWCOPPA
๐ŸŽ“ Start Learning๐Ÿ“‹ Parent Dashboard๐Ÿ  Command๐Ÿ—๏ธ Architecture๐Ÿค– 11โ†’42 Agents๐Ÿ’ฐ Compounder๐Ÿ”€ HybridFlow๐Ÿ›ก๏ธ Socratic Gate๐Ÿง  3-Layer Memory๐Ÿ—“๏ธ 9 Phasesโš–๏ธ Safety/Legal๐Ÿ’ต Costs๐Ÿง  Second Brain๐Ÿ‘ฅ Friends๐Ÿ“… Calendar๐Ÿ† Achievements๐Ÿ“– Timeline๐Ÿ“Š Year Review
HARDWARE
~$360
2x Pi5 8GB + peripherals
(Feb 2026 +$30 price increase)
MONTHLY
~$0
$0 local only
Cloud server pre-paid; optional T3
VS ALPHA
$40K+
Per child per year
($80K+ for both boys)
SERVICE COST BREAKDOWN
Ollama (qwen2.5:0.5b, 3b)$0Self-hosted, T0/T1 tiers
llama.cpp (Llama 3.1 8B)$0Self-hosted, T2 tier
Gemini 2.5 Pro (T3 cloud)~$1.25/$10 per 1M tokensInput/output, OFF by default
Neo4j Community$0Self-hosted Docker
PostgreSQL$0Self-hosted Docker
Vosk ASR + Kokoro TTS$0Self-hosted on Pi satellites
Cloudflare AI Gateway$0Cost tracking, logging, rate limiting (free plan)
Year 1 total: ~$360 (hardware only โ€” Pi satellites + peripherals). Cloud server is pre-paid and not a per-month cost for this project. Cloudflare AI Gateway provides cost tracking, request logging, and rate limiting at no cost (free plan). Cloud T3 is feature-flagged OFF by default. If enabled selectively, Gemini 2.5 Pro adds ~$1.25/1M input tokens, ~$10/1M output tokens. Raspberry Pi 5 prices increased Feb 2026 (+$30 for 8GB). Budget with margin and active cooling for sustained workloads.