🔍 Intelligence Synthesis JP/EN
Ethics · Regulation · Technology — Pharma Practice Notes

AI Tech Flash ── 2026-10-18

Google Releases Gemini 4 Argon to the Public

Google opened its newest model Oct. 13; prices double in December

What changes

Developers and businesses can use the model through an API, which is an interface that lets software talk to other software. Google describes it as built for long, multi-step work.

The introductory price ends Nov. 30. According to Tech Insider, input and output prices both double from Dec. 1. Teams testing it now should budget for production costs in advance.

Why it matters

According to PYMNTS, Google says the model ranks first on the Vals Index, a test of work in finance, coding, legal and tax. That is Google's claim. The PYMNTS article does not mention any independent check.

Google says security firm Wiz has already used the model to find and patch flaws that put patient records and government data at risk.

According to Tech Insider, the standard price will match Claude Opus 5.5 per token. Tokens are the small chunks of text AI companies use to meter usage.

What it is

Gemini 4 Argon is Google's newest flagship AI model. According to Tech Insider, Google moved it from limited preview to general release on Oct. 13.

Starsg's press release calls it Google's publicly announced frontier model for complex, long-horizon workflows. Here, a frontier model means the high-end class of model that Google labels "frontier."

Who and where

Google built it. According to PYMNTS, it opened Oct. 13 to developers, businesses and paying consumers.

According to Tech Insider, the API uses the model ID gemini-4-argon. It is also in Google AI Studio, Vertex AI and Google Antigravity, Google's coding environment.

Google AI Ultra subscribers get it in Antigravity. Ultra and Pro subscribers get it in the Gemini app and AI Mode in Search.

Tech Insider cites Google's API documentation: it is not available on the AI Pro plan and is excluded from the Free tier.

Why it was built

The articles give no direct statement of why Google built it. According to Starsg's press release, Google says it designed the model for demanding work such as software engineering, enterprise knowledge work and cybersecurity defense.

According to Tech Insider, Google also lists large code migrations and scientific research. The focus is long, autonomous work rather than quick single answers.

How it works

According to Tech Insider, the model has one shared limit of 1 million tokens for input and output combined. Earlier Gemini releases used separate limits for each.

The API has other caps. A non-streaming request is limited to 32,000 tokens, and a streaming call to 64,000.

Developers can choose a thinking level of low, medium or high. The setting trades response speed against depth of reasoning. The articles do not describe the model's inner workings.

Key points

  1. According to PYMNTS, Google opened Gemini 4 Argon on Oct. 13 to developers, businesses and paying consumers.
  2. According to Tech Insider, the introductory price is $2 per million input tokens and $10 for output. It becomes $4 and $20 on Dec. 1.
  3. Google says it ranks first on the Vals Index. The articles show no independent confirmation.

Conclusion

Google released Gemini 4 Argon on Oct. 13, aimed at long, complex work. The performance figures come from Google.

According to Tech Insider, the introductory price ends Nov. 30 and doubles from Dec. 1. Which surfaces you can use depends on your plan, so check before you rely on it.

Quotes and sources

“Google opened Gemini 4 Argon to developers, businesses and paying consumers on Tuesday (Oct. 13).”
── PYMNTS.com, 2026-10-15
“Google said it ranks first on the Vals Index, a test of work in finance, coding, legal and tax.”
── PYMNTS.com, 2026-10-15
“Gemini 4 Argon is Google’s publicly announced frontier model for complex, long-horizon workflows.”
── source, 2026-10-16
“During that window, input tokens cost $2 per million, output tokens cost $10 per million, and cached input tokens cost $0.10 per million, a 95% discount versus standard input pricing.”
── Tech Insider, 2026-10-17

Source articles

Written by AI (Claude) from the cited articles, then checked by Jev (TypeSafe's judgement model), a Gemini fact-check and a verbatim check of every quote.