Compute Economics
Fetching live data…

Inference ROI

Revenue per $1 API spend
—
BTC P&L ÷ API cost
Total API cost (30d)
—
all models
BTC P&L (30d)
—
import below ↓
Net margin
—
P&L minus API cost
API cost — daily (Anthropic live)
API cost BTC P&L ROI ×
By provider live

Trading floor — agent cost per job

Total agent cost today
—
all 6 agents
Sessions today
—
across all agents
Cache hit rate
—
avg across floor
Cost per session
—
blended avg
Agent breakdown — cost, cache & sessions
pending first report (23:55 UTC)
Sonnet 4.6 Haiku 4.5
Waiting for first report from agent_cost_audit.py — runs at 23:55 UTC daily
Agent cost trend — 7 days loading

Rate limit + caching — Anthropic

Peak tokens/min
—
— % of limit
Avg cache hit
—
30d
Rate limit
2M
tokens/min
Cache savings
—
tokens avoided
Rate limit use + caching — input tokens
Uncached tokens Cache hit rate % Rate limit

Loading…

Tokens per watt — energy efficiency

Tokens per watt-hour
—
input + output
Power draw
—
configured
Cost per M tokens
—
blended
Revenue per watt-hr
—
P&L ÷ energy
Power configurationfloor settings
Floor power draw (kW)
Hours/day active
Power cost ($/kWh)
Daily power cost
—
Monthly power cost
—
Total (API + power)
—

Model cost breakdown

Inference cost by model — 30d live
cost ROI

P&L data

Import BTC trading P&L
Paste daily CSV: 2026-06-01, 4820.50 · Negative for losing days
RC Quantum Automation · compute.economics.rcqsignals.com —