The same work on cloud APIs costs $5 and produces 4,320 grams of CO2. Below is live proof from our actual system.
0x
Cheaper than GPT-4o
0x
Cleaner carbon footprint
$0
Per 1,000 queries (CAD)
0%
Zero-emission power grid
1 Machine
14 services, zero cloud dependency
32 Cores, 48GB RAM
Handles everything locally
3 Sites Live
proaibuilder.com, oakdenproperty.com, oakden.ca
5+ AI Models
Local inference, under 500ms response
256-bit Encryption
All data encrypted at rest and in transit
94% Clean Grid
Ontario nuclear + hydro powering every query
Head to Head
Measured from live benchmarks on Oakden hardware. Ontario grid rates. No theoretical projections — real watts, real tokens, real cost.
Cost per query (CAD)
122x cheaper
CO2 per query
540x cheaper
Monthly cost (1,000 queries/day)
126x cheaper
24/7 always-on operating cost
1261x cheaper
Calculator
Drag the slider to match your expected AI usage. Watch the savings compound.
Cloud cost / month
$75.00
Oakden cost / month
$0.61
You save / year
$893
CO2 avoided / month
64.7 kg
A $3,000 Oakden Node pays for itself in 41 months at this usage level.
Provider Split
The Oakden Router classifies every query and routes it to the cheapest capable model. This is the real breakdown — and what it would cost if everything went to Opus.
Call distribution — 1 total queries
Token volume — 0.00M tokens processed
Local (Ollama)
—
Claude Haiku
—
Claude Sonnet
—
Claude Opus
—
If everything went to Opus
$0.00
CAD · all 0.00M tokens
—
cheaper with router
Actual cost with router
$0.0000
CAD · tracked spend
Cost breakdown by provider (CAD)
The Bridge
Oakden routes every request through local AI first. Only complex edge cases escalate to the cloud — and only when you allow it.
Your Query
AI Gateway
Complexity check
Local Node
31 tok/s · $0.000041/query
Private · Instant · No API key
Cloud Fallback
Only when needed
User-approved · Logged
Stays Local
May Escalate
You Control
Green AI
Ontario's power grid is 94% zero-emission — nuclear, hydro, wind, and solar. Every Oakden query runs on one of the cleanest grids on the planet.
Ontario Power Grid Composition
4.32g
CO2 per cloud query
US data center avg
0.008g
CO2 per Oakden query
Ontario clean grid
546x
Cleaner per query
Measured, not estimated
The Node
Dedicated hardware running multiple specialist AI models simultaneously. No GPU rental. No API keys. No monthly fees.
Reference Hardware
Active Models
proai-construction
LIVEConstruction domain specialist
Oakden AI
LIVEGeneral reasoning
All models run simultaneously on a single $3,000 machine. No cloud GPU required. No per-token billing.
Optimization Layers
Cost, carbon, and latency savings are compounding — not just one trick, but an entire stack optimized from hardware to model routing.
AI runs on your machine. Zero network round-trips. Zero per-token cost. 31+ tok/s sustained.
$0.000041 / query
AI gateway classifies each request. Standard stays local. Only premium-tier complexity touches cloud.
~95% stays local
Fine-tuned models (proai-construction) outperform generic cloud models on industry tasks with 1/100th the cost.
Custom-trained locally
Ontario's 94% zero-emission grid means every watt is clean. 25g CO2/kWh vs 400g US average.
546x cleaner
StudioIndex compresses 45K tokens of context to ~500 tokens per agent boot. Massive token savings at orchestration layer.
90x context reduction
Every query logs tokens, cost, energy draw, and CO2. Live dashboard proves savings are real — not marketing.
SQLite-backed proof
A $3,000 machine. Multiple specialist models. 1,000 queries for 4 cents. Powered by clean energy. No subscriptions. No per-token fees. That's the Oakden stack.