Today ainotis Join

My notis

ComputePublished Top storyAll news from that day

Ongoing case: What the AI buildout costs and needs 22 stories

Gemini 4 Argon used almost 25 trillion tokens in 72 hours, Pichai says

Google's chief executive told the Big Technology Podcast that Gemini 4 Argon's first 72 hours ran at about 3x Gemini 3.0's token use, and that Google is capacity constrained in serving it.

By ai notis. An AI newsroom with a human editor. Every fact links to its source. How we work

Share

Check our sources · 7 facts from 2 sources
Sundar Pichai: Path To Recursive Self-Improvement, Compute Allocation Dilemma, Future AI Interfaces: a video from Alex Kantrowitz on YouTubeVideo: Alex Kantrowitz on YouTube
watch on YouTube · Video still, linking to YouTube

Key points

  1. Pichai said Gemini 4 Argon had "almost up to 25 trillion tokens used" in its first 72 hours, about 3x what Gemini 3.0 had in the same period (02:43).
  2. He said of Argon's launch "we are capacity constrained as we're serving it" (02:57).
  3. He called Argon "one of one of the really good frontier models" across agentic work, knowledge work and coding, and said Google took longer with it (02:30).

What happened

Sundar Pichai says Google's new Argon model used "almost up to 25 trillion tokens" in its first 72 hours, "about 3x" what Gemini 3.0 had in the same period, and that Google is "capacity constrained as we're serving it" (02:43). Pichai is the chief executive of Google and Alphabet; he spoke to Alex Kantrowitz on the Big Technology Podcast, in a video published on 20 October 2026.

The figures are Pichai's own and Google has not been checked on them here. The automatic captions give "almost up to 25 trillion tokens used, which is about 3x uh what Gemini 3.0 had in the first 72 hours". Kantrowitz introduces the model as Gemini 4 Argon (01:16).

Pichai said of it that "across the board, be it for agentic work, knowledge work, coding" it is "one of one of the really good frontier models" (02:30), and that Google took longer with it than with other models given its capabilities (02:25).

On research, Pichai said "we do dedicate a meaningful percentage of our compute to these research bets" (41:08), naming synthetic data, continual learning and more efficient architectures as examples (40:58). He did not give the share.

Part of this interview is already in the record: his remarks on rollbacks after Argon (25 October) and on compute for the frontier and TPUs in space (21 October).

What it means for you

Our view

Pichai's figures point to strong early demand for Gemini 4 Argon, but they are his own and come with no breakdown of who used it or for what. His remark that Google is capacity constrained matters more to a buyer, because it suggests early limits on how much you can run.

That is our reading of what he said, not a statement of Google's terms. If you plan to build on Argon, check the rate limits and quotas on your account and test your expected load before you move a production workload.

It adds no new facts.

Share this story

Your reaction

We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.

Check our sources

Every sentence above is checked against these 2 sources.

1 Google CEO Sundar Pichai: AI Infinite Possibilities, Compute Scarcity Dilemmas, Future AI ProductsBig Technology Podcast (YouTube) · 6 facts Open the source archived copy

  1. Pichai said that in the first 72 hours since launch Argon had "almost up to 25 trillion tokens used, which is about 3x uh what Gemini 3.0 had in the first 72 hours" (02:43). These are the automatic captions' words.

    (02:43) And you know, in the first 72 hours since we launched, it's you know, almost up to 25 trillion tokens used, which is about 3x uh what Gemini 3.0 had in the first 72 hours.
    cite

  2. Pichai said "we are capacity constrained as we're serving it" about Argon's launch (02:57).

    (02:57) So, uh and we are capacity constrained as we're serving it.
    cite

  3. Alex Kantrowitz introduced the model as "a Gemini 4 Argon" in his opening question (01:16).

    (01:14) So, Google is back on the frontier with a Gemini 4 Argon.
    cite

  4. Pichai said of Argon "across the board, be it for agentic work, knowledge work, coding" it is "one of one of the really good frontier models" (02:30), and that Google "took a bit longer than our other models given given its capabilities" (02:25).

    (02:25) Uh we took a bit longer than our other models given given its capabilities. But I think across the board, be it for agentic work, knowledge work, coding, uh creative tasks, uh I think Argon is uh you know, one of one of the really good frontier models.
    cite

  5. Pichai said "we do dedicate a meaningful percentage of our compute to these research bets, too" (41:08); he did not give the percentage in that answer.

    (41:08) So, part of we do dedicate a meaningful percentage of our compute to these research bets, too.
    cite

  6. The video was published on the Big Technology Podcast YouTube channel on 20 October 2026, titled "Google CEO Sundar Pichai: AI's Infinite Possibilities, Compute Scarcity Dilemmas, Future AI Products"; Alex Kantrowitz is the interviewer.

    Watch page metadata: publishDate 2026-10-20T10:41:54-07:00; title: Google CEO Sundar Pichai: AI's Infinite Possibilities, Compute Scarcity Dilemmas, Future AI Products
    cite
2 Sundar Pichai, Google Blog author pageGoogle · 1 fact Open the source archived copy

  1. Sundar Pichai is CEO of Google and Alphabet, per Google's blog author page.

    cite

Topics

The morning email

On the mornings we publish: the three top stories and up to four short ones. Free.

We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep