The news
Google on October 13, 2026 opened Gemini Argon, formally Gemini 4 Argon, to developers, enterprises and paying consumers, giving businesses its newest flagship AI model through the Gemini API, AI Studio and its business plans. The model is trending in US searches as buyers size it up against rivals.
The wider release follows a September 30 announcement in which Google limited access to vetted cyber defenders in its Fairwind Program, according to AI Pricing Guru, which reported at the time that no public date had been given. Google's October 13 post, written by Koray Kavukcuoglu, senior vice president at Google DeepMind and Google's chief AI architect, said the model is now globally available to developers and enterprises.
Google is pitching Argon at long, multi-step work in coding, finance, legal, cybersecurity and science. Through the Gemini API it offers a combined input and output limit of 1 million tokens, the units of text that models read and write. Google said Argon ranked first on the Vals Index, which measures finance, coding, legal and tax tasks, and that attacks succeeded only 0.7% of the time on the Gray Swan indirect prompt injection benchmark, a test of whether hidden instructions can hijack a model.
Pricing is introductory. Through November 30, 2026, the API costs $2 per million input tokens and $10 per million output tokens, with cached input tokens 95% off. After that date the public model costs $4 and $20 respectively, according to Google.
Access varies by plan. Google AI Ultra subscribers get Argon in Google Antigravity, the company's agent coding tool, and in AI Studio, while Ultra and Pro subscribers get it in the Gemini app and AI Mode in Search. Enterprises can use it through Gemini for business, the API and Antigravity. A separate Google developers post said the API model ID is gemini-4-argon and that Argon scored 87% on the ERQA embodied reasoning benchmark used in robotics.
Google also cited internal results: agentic code submissions in Antigravity rose more than 35% and code rollbacks fell by nearly a third, the company said. These are Google's own figures and have not been independently verified.
The numbers
- Intro API price (to Nov 30)
- $2 in / $10 out per 1M tokens
- Price after Nov 30
- $4 in / $20 out per 1M tokens
- Cached input discount
- 95%
- Token limit (Gemini API)
- 1M unified input-output
- Prompt-injection attack success (Gray Swan)
- 0.7%
Why CEOs should care
For technology buyers, the introductory window is the most practical part of this launch. At $2 and $10 per million tokens until November 30, teams can run Argon against their current model on real workloads at half the price Google will charge afterward. Build the comparison on the post-November price, not the promo, so a pilot does not turn into a budget surprise in December.
CFOs and procurement leads should treat a new flagship from a major vendor as leverage. Ask incumbent providers how their per-token and cached-token rates compare with Google's published list prices, whether committed-spend contracts allow switching models, and what notice they give before price changes. Contracts that lock in one model family for years deserve a second look.
CISOs have a specific reason to test. Google says Argon cut indirect prompt injection success to 0.7% on one benchmark, a risk that grows as agents read emails, web pages and documents. Vendor benchmarks are a starting point, so ask for results on your own red-team cases before giving any agent access to production systems.
The bigger picture
Google started Argon with a restricted, security-first rollout before widening access, an approach that lets a vendor gather feedback on misuse before broad release. The broader release puts a new top-tier model on the market with an explicit price step-up, which gives buyers a clear deadline for evaluation and gives competitors a public price point to answer.
Google is also offering 10,000 scientists the chance to apply for six months of free access to Antigravity's highest usage tier, and it is expanding the Fairwind Program to hundreds of additional partners, the company said. Both moves aim to widen use of the model in research and security work beyond paying API customers.
What’s next
Watch whether rival model makers adjust list prices before Google's introductory rate ends on November 30, 2026, and whether Google confirms a separate price for very long prompts. Enterprises should finish benchmark tests and contract reviews before the higher $4 and $20 rates take effect.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story
