COMPUTE
. FINANCE

Oracle.

Compute Price Index

CPI

A weighted reference price for AI compute, derived from twelve models across four providers.

1 SCU — STANDARD COMPUTE UNIT

$0.007245

Cost of a reference workload — 1,000 input tokens + 500 output tokens — averaged across the basket. Updated 19h ago.

FRONTIER (30%)

$0.004199

4 models · capped mean contribution

STANDARD (40%)

$0.002505

4 models · capped mean contribution

LIGHTWEIGHT (30%)

$0.00054

4 models · capped mean contribution

Provider share by model count

OpenAI 25%Anthropic 25%Google 25%xAI 25%

The basket.

One model per provider per tier. Twelve models total. Current basket: v1.0 (launched April 7, 2026). Index history runs back to v0 inception on June 17, 2025.

Frontier Tier

30% weight
Frontier Tier model prices, USD per million tokens, 5% routing fee included
ModelProviderReleasedInput $/1M Output $/1M
Grok 4xAI2026-03$3 ($3.15)$15 ($15.75)
Claude Opus 4.7Anthropic2026-04$5 ($5.25)$25 ($26.25)
GPT 5.5OpenAI2026-04$5 ($5.25)$30 ($31.5)
Gemini 3.1 ProGoogle2026-02$2 ($2.1)$12 ($12.6)

Standard Tier

40% weight
Standard Tier model prices, USD per million tokens, 5% routing fee included
ModelProviderReleasedInput $/1M Output $/1M
Claude Sonnet 4.6Anthropic2026-02$3 ($3.15)$15 ($15.75)
GPT-4.1OpenAI2025-04$2 ($2.1)$8 ($8.4)
Gemini 3 ProGoogle2025-11$2 ($2.1)$12 ($12.6)
Grok 3 MinixAI2025-02$0.3 ($0.31)$0.5 ($0.53)

Lightweight Tier

30% weight
Lightweight Tier model prices, USD per million tokens, 5% routing fee included
ModelProviderReleasedInput $/1M Output $/1M
Claude Haiku 4.5Anthropic2025-10$1 ($1.05)$5 ($5.25)
GPT-5 MiniOpenAI2026-01$0.25 ($0.26)$2 ($2.1)
Grok 4.1 FastxAI2026-01$0.2 ($0.21)$0.5 ($0.53)
Gemini 3 FlashGoogle2025-12$0.5 ($0.53)$3 ($3.15)

SCU History

Hourly SCU history. Each step is a new on-chain version; values are constant between updates.

Total SCUFrontierStandardLightweight
Jun 25Aug 25Sep 25Nov 25Dec 25Jan 26Feb 26Mar 26May 26$0.0002$0.0027$0.0052$0.0085

Current

$0.00724

Period Change

+11.92%

High / Low

$0.00822 / $0.00647

Reconstitutions

10

Display currencies. The canonical SCU is denominated in USD and stored on-chain as USDC microcents. Non-USD values shown above are display-only conversions sourced from a public FX feed. They are not part of the on-chain CPI value and may differ slightly from any specific FX provider.

Compute Power Index

Purchasing power per dollar — higher is better. Anchored to base 100 at the v0 inception snapshot (June 17, 2025).

A reading of 120 means compute is 20% cheaper than baseline. The Compute Power Index is the inverse of the SCU, normalized — when costs go down, the line goes up.

Compute Power IndexFrontierStandardLightweight
Jun 17, 25Jul 29, 25Sep 8, 25Oct 20, 25Dec 1, 25Jan 11, 26Feb 23, 26May 8, 26376287136

Current Index

89.4

Period Change

-10.6

High / Low

100.0 / 78.7

Baseline

100.0

Reconstitution Log

Every basket version update is recorded on-chain with a timestamp, the affected models, and the resulting SCU. The log spans v0 (June 2025 — April 2026) and v1.0 (the public launch basket).

v10May 7, 20262 changes

Added gpt-5.5; Removed gpt-5.4

SCU: $0.006494$0.007244(+11.5%)

v9Apr 16, 20262 changes

Added claude-opus-4.7; Removed claude-opus-4.6

SCU: $0.006494$0.006494(+0.0%)

v8Mar 1, 20266 changes

Added claude-sonnet-4.6, grok-4, claude-opus-4.6; Removed claude-3.5-sonnet, claude-3-opus, grok-3

SCU: $0.00822$0.006494(-21.0%)

v7Feb 19, 20262 changes

Added gemini-3.1-pro; Removed gemini-2.5-pro

SCU: $0.008023$0.00822(+2.5%)

v6Feb 1, 20262 changes

Added gpt-5.4; Removed gpt-4o

SCU: $0.007742$0.008023(+3.6%)

v5Jan 1, 20264 changes

Added gpt-5-mini, grok-4.1-fast; Removed grok-3-fast, gpt-4o-mini

SCU: $0.007629$0.007742(+1.5%)

v4Dec 17, 20252 changes

Added gemini-3-flash; Removed gemini-2.0-flash-lite

SCU: $0.007429$0.007629(+2.7%)

v3Dec 1, 20252 changes

Added gemini-3-pro; Removed gemini-2.5-flash

SCU: $0.006499$0.007429(+14.3%)

v2Oct 1, 20252 changes

Added claude-haiku-4.5; Removed claude-3.5-haiku

SCU: $0.006473$0.006499(+0.4%)

v1Jun 17, 202514 changes

Inception with 12 models

Methodology

The SCU is calculated from public pricing data using a fixed, deterministic formula. Anyone can reproduce it.

Step 1 — Per-model cost

For each model in the basket, compute the cost of the reference workload using the provider's published per-token pricing.

cost_m = (input_price_m × 1000) + (output_price_m × 500)

Step 2 — Per-tier capped mean

For each tier t ∈ {frontier, standard, lightweight}, compute the arithmetic mean of all four model costs in that tier. Any single model whose cost exceeds 2× the tier mean is capped at 2× — preventing one expensive model from dominating the tier average.

tier_avg_t = mean(min(cost_m, 2 × tier_mean_t))

Step 3 — Weighted SCU

Apply tier weights and sum: Frontier 30%, Standard 40%, Lightweight 30%. The weights reflect real-world usage distribution across capability tiers.

SCU = (frontier_avg × 0.30) + (standard_avg × 0.40) + (lightweight_avg × 0.30)

On-Chain Verification

The basket and SCU live in the OracleRegistry contract on Base. Each version is a single transaction; the on-chain payload is canonical.

NetworkBase (Chain ID 8453)
Index inceptionJune 17, 2025 (basket v0)
Current versionv10 (live)
Last updateMay 7, 2026 at 03:41 PM UTC

API Reference

Public read endpoints. No authentication required. Full reference in the documentation.

VIEW DOCS →
1 SCU = $0.00724·Claude Haiku 4.5 $1 / $5+5%Claude Sonnet 4.6 $3 / $15+5%GPT-4.1 $2 / $8+5%GPT-5 Mini $0.25 / $2+5%