Gemini 4 Argon reaches general availability after two weeks of staged rollout

Google marked Gemini 4 Argon as generally available in the Gemini API changelog dated October 13, 2026, listing it under the model id gemini-4-argon as its most capable model for complex coding, agentic, math and science tasks. The launch page now frames availability as "for everyone," but The Verge reports consumer access is currently limited to higher-tier Pro and Ultra subscribers, with no stated timeline for Plus or free users. General availability here means a production API path, not an unlocked app tier.

The changelog entry makes it official

LinkLoot covered the October 8 announcement, when Argon was reachable only through Google's Fairwind Program for trusted cyber defenders, with broader paid access promised later. The changelog moves the model across that line: the gemini-4-argon id is listed as generally available, and the entry points to a "What's new in Gemini 4 Argon" section covering API changes, pricing and migration guidance. That is the practical difference between the staged preview and today's state: developers can build against a stable, non-preview model string.

What the API actually gives you

The changelog describes Argon as Google's most capable model for coding, agentic workflows, math and science. Google's announcement claims up to 1 million output tokens, with vendor-reported results of 77.9% on DeepSWE v1.1 and 51.3% on Zapier's AutomationBench. The Verge adds that Google positions the model against OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 and Opus 5.5 across coding, cybersecurity, science, finance and law benchmarks. Every one of those numbers comes from Google; none has been independently reproduced at publication time, so treat them as positioning, not guarantees.

App rollout stays tier-gated

The Verge's launch reporting, which matches the radar lead detected for the official blog post, puts Argon in the Gemini app, AI mode in Search, and Google's agentic development platform Antigravity for Pro and Ultra subscribers, following the same trusted-defender preview that preceded the announcement. Google's launch page now headlines availability "for everyone," but The Verge's concrete subscription breakdown shows Plus and free availability remains open. If you are deciding whether an Argon feature is reachable in your plan, check your own app tier and AI Studio model selector rather than the headline.

Pricing holds, with a November step-up

API pricing matches what Google introduced at announcement time: $2 per million input tokens and $10 per million output tokens, with cached input discounted 95% from the input price. The Verge notes the introductory rate doubles at the end of November, which still leaves per-token costs well below rival frontier models on paper. Per-token price is not total cost, though: long-horizon agentic runs can consume large numbers of tokens for one task, and Google has not published comparable per-task cost data.

Why teams are paying attention

Argon is Google's first major frontier release since Gemini 3 in late 2025, and The Verge frames it as Google re-entering a race it had trailed for months, landing a day after Meta announced its own Muse Wonder Full model. Google also describes internal deployments: vulnerability discovery and patching, halving the compute behind YouTube's comments section, and debugging a Wing drone-delivery physics simulator. Those accounts come from Google and remain unverified externally.

Before you switch production workloads from Gemini 3.8 Flash or a rival model, run Argon through your own coding and agent evals, model real token consumption at the doubled post-November rate, and confirm which surfaces actually accept your plan tier. The full capability matrix lives in Google's "What's new in Gemini 4 Argon" documentation and its model pages, which remain the reference as the rollout completes. For workflow ideas beyond raw benchmarks, see our guide to AI agent tools.

Source note: availability status comes from the official Gemini API changelog and Google's launch page, read directly; rollout tiers, benchmark comparisons and pricing context were cross-checked against The Verge's independent reporting. Claims such as benchmark results, internal deployments and introductory pricing are vendor-reported and have not been independently tested by LinkLoot.