Otto365 AI Brief #003

Gemini: Google's new model targets legal and finance work, with a catch

How this was made: Otto365 is an animated character and his voice is generated by AI. The video tells the same story as the text on this page; the transcript lists every spoken word and the main lines shown on screen. Spotted an error?
  • Published
  • By the Otto365 AI Brief. Checked and approved by Peter Ferguson, Editor, on . The script was drafted with AI assistance, then every sentence was checked against the sources below.
  • Announced 30 Sep 2026
  • Voiced by AI

What happened

Google's new Gemini model targets legal, finance and coding work, with a catch. Google DeepMind announced Gemini 4 Argon, its new frontier model, on the 30th of September 2026. It is built for long, multi-step work in coding, legal, finance and cyber defence.12

Three things to know

  1. Long jobs in one go First, long jobs in one go. Google DeepMind says the output limit rises to a million tokens, up from 64 thousand, and it tops Zapier's AutomationBench for business tasks at 51.3 percent.1
  2. Finds and patches flaws Second, cyber defence. Google DeepMind says Argon can find, validate and patch critical software vulnerabilities autonomously. At Wiz, it uncovered a flaw exposing personal information in hospital software worldwide that previous models missed.1
  3. Access and price Third, access. Only trusted cyber defenders in Google's Fairwind Program have it. Paid API customers and Google AI Ultra subscribers follow, at a starting price of 2 dollars per million input tokens and 10 per million output.1

Why it matters to you

Otto365's take (opinion)

So, my take. Google is gating its new model behind safety testing, but the price is set, and its targets, legal drafting, financial research and debugging, are what UK small firms buy by the hour.12

Do this week

Try this: list one long, multi-step task you outsource.

Sources

  1. Google DeepMind (opens in a new tab) Primary source, accessed 1 Oct 2026
  2. The Verge (opens in a new tab) News report, accessed 1 Oct 2026
The quotes we checked (17)

  • Gemini 4 Argon: our next era of frontier intelligence Sep 30, 2026

    Google DeepMind · source 1

  • Sep 30, 2026, 8:41 PM UTC Illustration: The Verge

    The Verge · source 2

  • Today, we’re announcing our new frontier model, Gemini 4 Argon, which is rolling out to a set of trusted cyber defenders through our Fairwind Program.

    Google DeepMind · source 1

  • It delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.

    Google DeepMind · source 1

  • Built to sustain deep reasoning across complex, long-horizon workflows

    Google DeepMind · source 1

  • we are significantly expanding the model’s output token limit to an industry-leading 1M tokens, up from the previous 64K tokens.

    Google DeepMind · source 1

  • When the model has the headroom to think deeply and generate hundreds of thousands of tokens in a single trajectory, it adds a new level of depth in reasoning to solve tough problems in one go.

    Google DeepMind · source 1

  • On AutomationBench, Zapier’s benchmark measuring end-to-end execution across core business functions, Argon ranks #1 with a score of 51.3%.

    Google DeepMind · source 1

  • Argon can autonomously find, validate, and patch critical software vulnerabilities.

    Google DeepMind · source 1

  • the model uncovered a critical vulnerability exposing sensitive personal information across healthcare software used by hospitals worldwide, identifying a severe risk that previous frontier models had missed.

    Google DeepMind · source 1

  • Wiz is already using Argon for cybersecurity defense through its Scan for Good initiative - a program dedicated to protecting critical public infrastructure for free by finding and remediating high-risk exposures.

    Google DeepMind · source 1

  • before we release to developers, enterprises, and consumers, starting with paid API customers and Google AI Ultra subscribers.

    Google DeepMind · source 1

  • of $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price.

    Google DeepMind · source 1

  • Safely releasing frontier capabilities at this level requires a phased approach.

    Google DeepMind · source 1

  • Vals Finance Agent v2 (multi-step financial research) and Harvey’s Legal Agent Benchmark (legal research and drafting).

    Google DeepMind · source 1

  • Google engineers have been using Argon for their daily tasks, from everyday debugging to large-scale codebase migrations and algorithm designs.

    Google DeepMind · source 1

  • Gemini 4 Argon is launching first in a limited capacity so Google can make sure it’s not misaligned.

    The Verge · source 2

Transcript

Full transcript

Google's new Gemini model targets legal, finance and coding work, with a catch.

Play from 0:06 Google DeepMind announced Gemini 4 Argon, its new frontier model, on the 30th of September 2026. It is built for long, multi-step work in coding, legal, finance and cyber defence.

Play from 0:19 First, long jobs in one go. Google DeepMind says the output limit rises to a million tokens, up from 64 thousand, and it tops Zapier's AutomationBench for business tasks at 51.3 percent.

Play from 0:33 Second, cyber defence. Google DeepMind says Argon can find, validate and patch critical software vulnerabilities autonomously. At Wiz, it uncovered a flaw exposing personal information in hospital software worldwide that previous models missed.

Play from 0:50 Third, access. Only trusted cyber defenders in Google's Fairwind Program have it. Paid API customers and Google AI Ultra subscribers follow, at a starting price of 2 dollars per million input tokens and 10 per million output.

Play from 1:06 So, my take. Google is gating its new model behind safety testing, but the price is set, and its targets, legal drafting, financial research and debugging, are what UK small firms buy by the hour.

Play from 1:19 Try this: list one long, multi-step task you outsource.

I'm Otto365.

Follow the Brief.

On screen

  • Google's new Gemini model legal, finance and coding
  • Announced 30 Sep 2026
  • 1. Long jobs in one go
  • 2. Finds and patches flaws
  • 3. Access and price
  • Gated, but priced
  • Try this: List one multi-step task
  • Source: Google DeepMind, 30 Sep 2026
  • Voiced by AI. Checked by humans. Apps 365 sells AI and Microsoft 365 services. Not endorsed by companies named.

Voiced by AI. Checked by humans. Apps 365 sells AI and Microsoft 365 services. Not endorsed by companies named. Report an error.