Jarvis V2

v2.0.0
0/7 Agents
Uptime:
BQ
API
Live
Loading market data...

CapEx & OpEx — System Financials

Variable and fixed costs of running the Shanfeld Equities AI intelligence engine

Monthly OpEx

$256.50

All-in operating cost

Annual Run Rate

$3,078.00

Projected yearly

API Costs

$187.00

73% of total

Infrastructure

$69.50

27% of total

Cost / AUM

0.0 bps

Annual cost basis

API Costs by Agent — Variable OpEx

Claude Opus 4.6Anthropic
$97.50/mo
~50 calls/day
Input: $15/1M tokOutput: $75/1M tok~3.3M tok/mo
o1OpenAI
$33.00/mo
~30 calls/day
Input: $15/1M tokOutput: $60/1M tok~1.3M tok/mo
Claude Sonnet 4.5Anthropic
$27.00/mo
~80 calls/day
Input: $3/1M tokOutput: $15/1M tok~4.2M tok/mo
Grok 3xAI
$16.50/mo
~60 calls/day
Input: $3/1M tokOutput: $15/1M tok~2.7M tok/mo
Sonar-ProPerplexity
$12.00/mo
~40 calls/day
Input: $3/1M tokOutput: $15/1M tok~2.0M tok/mo
Gemini 2.0 FlashGoogle
$1.00/mo
~100 calls/day
Input: $0.1/1M tokOutput: $0.4/1M tok~5.5M tok/mo
Total API OpEx$187.00/mo

Infrastructure — Fixed OpEx

Vercel Pro

Dashboard hosting + CI/CD

$20.00/mo

Hosting

Cloudflare Tunnel

Free tier — tunnel to Mac Studio

FREE

Networking

BigQuery

Decision logging + analytics (estimated)

$15.00/mo

Data

Google Drive API

Fidelity portfolio sync

FREE

Data

Domain + SSL

shanfeldequities.com

$1.50/mo

Infrastructure

Mac Studio (electricity)

24/7 operation — M2 Ultra estimated

$25.00/mo

Hardware

Synology DS225+ (optional)

NAS storage — 2x 4TB Seagate Barracuda

$8.00/mo

Hardware

Total Infrastructure$69.50/mo

Cost Optimization Opportunities

Use Sonnet for Routing$25/mo

Route simple queries through Claude Sonnet 4.5 ($3/M) instead of Opus ($15/M). Saves ~$25/mo for non-strategic queries.

Difficulty:Easy
Gemini for Bulk Analysis$40/mo

Gemini 2.0 Flash at $0.10/M input is 150x cheaper than Opus. Use for high-volume quantitative scans.

Difficulty:Easy
Cache Repeated Queries$50/mo

Implement semantic caching for similar questions within a session. Reduces API calls by ~30%.

Difficulty:Medium
Prompt Compression$15/mo

Reduce system prompt sizes across agents. Each 1K token saved across 300 daily calls = significant savings.

Difficulty:Medium
Batch Processing$35/mo

Use Anthropic Batch API (50% discount) for non-real-time analysis: end-of-day reviews, portfolio rebalancing.

Difficulty:Medium
Local Model for Triage$30/mo

Run a small local model (Llama 3) for query classification before routing to paid APIs. Reduces unnecessary API calls.

Difficulty:Hard

ROI Analysis

Breakeven Alpha

0 bps

Annual alpha needed to cover system costs

Cost per Decision

$0.12

Avg cost per consensus analysis (~50/day)

Cost per $1K Managed

$3078000.00

Annual cost per $1K of portfolio

vs Hedge Fund Fees

99% cheaper

2/20 hedge fund = ~$0/mo vs $256.50

Note: API costs are estimates based on projected usage. Actual costs vary based on query complexity, response length, and daily volume. All costs shown assume the current paper trading configuration. Production trading may increase API usage 2-3x.