Skip to content
Palanor is in early release, with our founding customers.Become a founding customer →
PalanorPalanor

Line

U.S. vs. China: Capability of the Best AI Model

Highest Epoch Capabilities Index score reached by a model from each country’s developers, monthly

Best U.S. model: 167 in Oct 2026+31 since Oct 2024

Plays the chart through its history. Pauses are where the data set a record or moved most.

Epoch AI (CC BY 4.0). A model counts for a country when one of its developers is based there. Each line is the best score among models released by that month; it rises when a new model beats the old best. From January 2024, when Epoch had scored enough Chinese models to show a frontier.

The Epoch Capabilities Index puts every model on one scale built from about forty benchmarks. The two lines trace the best model each country has released: the U.S. line has stayed on top, and the distance between them is the race.

What this chart reads

Sources

United States · Best AI Model Capability (Epoch Capabilities Index). Source: Epoch AI, Epoch Capabilities Index (epoch.ai/eci). CC BY 4.0. Rewritten weekly, because Epoch re-estimates every score when it adds benchmarks.

China · Best AI Model Capability (Epoch Capabilities Index). Source: Epoch AI, Epoch Capabilities Index (epoch.ai/eci). CC BY 4.0. Rewritten weekly, because Epoch re-estimates every score when it adds benchmarks.

Charts are drawn from Palanor's signal library and recomputed from source data on every view. Transformations (changes, indexing, year-on-year rates) are Palanor's.

Related charts

AI & technologyForecast

U.S. Businesses Using AI

Share of U.S. businesses using AI to produce goods or services, and share expecting to within six months, %

23.8%Data 33 days ago
AI & technologyForecastMoving

AI in Text Index: S&P 500 Earnings Calls

How often calls use the words language models overuse, against their level before ChatGPT (January 2021 to September 2022 = one), trailing 90 days

1.05Data today
AI & technologyForecast

What Companies Talk About on Earnings Calls

Average mentions per S&P 500 earnings call, rolling 90 days

11.41 viewsData today
AI & technology

Hyperscaler Capital Spending

Capital expenditures, trailing 12 months, $ billions

$173BData to Jun 2026
AI & technologyForecastMoving

AI in Text Index: 10-K Risk Factors

Words language models overuse in S&P 500 annual reports, Item 1A, latest 10-K per company, against the level in 10-Ks filed before ChatGPT (= one)

1.41Data today
AI & technologyForecast

AI in Text Index: Earnings Calls by Sector

Words language models overuse, per 10,000 words of S&P 500 earnings calls, trailing 90 days, sectors with enough companies to read

5.9Data today

Cookies

We only use essential cookies to keep you signed in and remember your theme. No analytics, no tracking, no ads. See the Privacy Policy for the full read.