Google launches workplace Gemini agent and expands access to Argon model

Google Cloud introduced Gemini agent on October 8 as a product for corporate users at its Gemini at Work 2026 conference, according to 3DNews. Google Cloud CEO Thomas Kurian called it "your new single, universal agent for work," The Register reported.

The pitch goes beyond answering questions. According to TechCrunch, the agent can plan and execute tasks across business applications and systems, delegate work to subagents and use multiple AI models. CBS Top Stories reported that Google said it could write and run code and integrate into applications such as Gmail.

The October 8 coverage also described how Google would place the agent inside a workplace. The Next Web reported that each agent gets its own Workspace account, email address, calendar and company directory entry. It can continue working for days after a laptop closes, with every action logged.

Access was initially limited. AI on vc.ru reported on October 8 that the agent was available through a subscription in closed preliminary testing. That is separate from the later, broader release of the Gemini 4 Argon model.

On October 11, TestingCatalog reported that users on X were seeing Gemini 4 Argon in A/B tests on Antigravity. Early outputs reportedly showed sharper interface design than Gemini 3.8 Flash, but those observations were not a general release or an independent performance assessment.

On October 12, TestingCatalog reported references to GemPix 3 Pro suggesting that Google was developing an Argon-based image model that could become Nano Banana 3 Pro. The report described signs of development, not a confirmed launch.

The broader Argon rollout arrived on October 13. 9to5Google reported that the model was starting to reach the Gemini app and Google Antigravity following an announcement at the end of September. Business Insider reported Google's claims that Argon excels in coding and knowledge work in finance and law.

Alongside that release, Google integrated Antigravity into the Gemini app for macOS, according to 9to5Mac. That update extended the rollout beyond access to the model itself.

Gemini 4 Argon is now available to developers, businesses and paid users. Digital Trends reported global availability through Google's paid API tier, Antigravity, the Gemini app and AI Mode in Search. 9to5Google identified AI Pro and AI Ultra as the subscription plans receiving the Gemini app rollout.

For API customers, launch pricing is $2 per million input tokens and $10 per million output tokens, according to The Next Web. The outlet reported that these are discounted rates, compared with standard prices of $4 and $20 respectively.

The workplace agent and Argon model are related developments, but their availability should not be confused. The supplied coverage establishes broader paid access to Argon; it does not establish that the enterprise agent has moved beyond its initial closed testing.

The next practical issue is whether Google expands access to the enterprise agent beyond closed testing, and on what subscription terms. The reports describe substantial autonomy, including background work and delegation, but do not establish how reliably those features perform across ordinary business tasks.

Also pending is whether the image-model references reported by TestingCatalog become an announced product. For now, they remain indications of development rather than another available Argon service.