Interconnecting Intelligence The facts on AI, from demand to buildout.

How many tokens do Google's AI models process?

Google's model APIs went from 7 billion tokens a minute to nearly 32 billion.

Updated by hand when the source publishes

Google's model APIs went from 7 billion tokens a minute to nearly 32 billion. Tokens processed per minute through Google's first-party model APIs, by quarter, as stated by Alphabet Data in the table below.
Download image Point at the chart, or tap it, to see the values. The image is sized for X and LinkedIn, with source and address. Sources: [2] [4] [5] [6]
Show the data
QuarterTokens per minute, as stated
Jul to Sep 20257 billion
Oct to Dec 2025over 10 billion
Jan to Mar 202616 billion
Apr to Jun 2026approximately 22 billion
Jul to Sep 2026nearly 32 billion

Key figures

Tokens a minute, billions, Q3 2026
nearly 32
Tokens a minute, billions, Q3 2025
7

Worked out from the chart's data.

How it's measured

Alphabet states in its quarterly earnings remarks how many tokens its first-party models process through its APIs, most recently as tokens per minute. Each bar is the figure as stated, with Alphabet's own wording ('over', 'about', 'nearly') kept.

This counts API traffic to Google's own models, not every Google product. The figures are entered by hand each quarter from Alphabet's remarks.

Sources

The data

Download as CSV ยท Free to use with credit to Interconnecting Intelligence and the original source.

QuarterTokens per minute, as stated
Jul to Sep 20257 billion
Oct to Dec 2025over 10 billion
Jan to Mar 202616 billion
Apr to Jun 2026approximately 22 billion
Jul to Sep 2026nearly 32 billion

More Demand charts

  1. Tokens by lab on OpenRouter
  2. What AI spending goes to
  3. Largest apps on OpenRouter
  4. Which chatbots Americans use
  5. AI assistant users
  6. Businesses using AI

All Demand charts

Latest Demand facts

See every Demand fact