$CPI • COMPUTE CPI
100.0
MoM:
—
YoY:
—
Base = 100 (February 2026)
$JUDGE
—
Judgment CPI
$LCTX
—
LongContext CPI
$BULK
—
Budget Tier
$FRONT
—
Frontier Tier
Spreads
$COG-P
Cognition Premium
—
$JDG-P
Judgment Premium
—
$CTX-P
Context Premium
—
Methodology
Compute CPI tracks the changing cost of a representative "basket" of AI work, expressed as an index where February 2026 = 100.
The basket includes six workload categories: chat/drafting, summarization, classification, coding, judgment/reasoning, and long-context synthesis. Each category is weighted to reflect typical institutional usage patterns.
View full methodology →No referral fees. No sponsored rankings.
Public APIs. Auditable methodology.
Data sources: OpenRouter (prices), Artificial Analysis (quality grades), simonw/llm-prices (historical)
Coming Q2 2026
- Civic CPI (public sector workload basket)
- Quarterly reports with trend analysis
- Historical data explorer
- API access for researchers