Google Opens Gemini 4 Argon to Developers and Paying Subscribers

Google began letting developers build with Gemini 4 Argon on October 13, 2026, positioning the model as its answer to long-horizon software work: code migrations, frontier research, enterprise knowledge tasks and cybersecurity defense. The newsroom post describes Argon as handling multi-step problems with ease, a claim aimed at engineers who have spent months stitching together smaller models to cover the same ground. Access is split by payment tier. Developers reach Argon through the paid tier of the Gemini API inside Google AI Studio and Android Studio, while Google AI Ultra subscribers, the company’s priciest consumer plan, get it inside AI Studio itself and inside Antigravity, Google’s agent-first development platform aimed at teams that want an AI system managing its own build and test loop rather than a person running each step by hand.

The release follows weeks of guarded language from Google about Argon’s capabilities. A companion post from the safety and security team, linked beside the main announcement, frames the rollout as deliberately staged because of the model’s potential for misuse, a line several outlets covering the story returned to even though Google’s own developer-facing post barely touches the subject, leaving the caution almost entirely to the parallel safety post rather than the page developers actually read when they sit down to build.

Google is also steering Argon toward a specific audience rather than a general refresh. The newsroom post links guides for Android Studio, for Antigravity and for robotics use cases, suggesting the company wants Argon read as infrastructure for agentic coding tools rather than a chat model update. That framing puts Argon closer to Google’s developer product line than to the consumer Gemini app, even though the same model eventually reaches both surfaces.

The numbers

The announcement names four separate entry points for the model: the paid Gemini API tier, Google AI Studio, Android Studio and the Antigravity platform, each pointed to its own guide rather than one consolidated page. Google AI Ultra is named as the one subscription plan that gets direct AI Studio access alongside developer accounts, which is the detail missing most from Google’s own post: no stated per-token price, no context window figure and no mention of what, if anything, separates Argon’s paid-tier limits from the model it replaces. Readers tracking cost have to go to Google’s pricing page rather than this announcement, a gap that is unusual for a vendor newsroom post and notable mainly for what it leaves out.

Teams already paying for Gemini access should expect the usual pattern from Google: a capability shipped first, with allowances and prices published later on a separate page. Until that page updates, anyone budgeting around Argon for a production coding workflow is working from the older rate card, not one built for this model.