Gemini 4 Argon is generally available: gemini-4-argon ships in the Gemini API and AI Studio with a 1M unified window, long decode and $2/$10 intro pricing

Gemini 4 Argon, announced on Sep 30 for Fairwind cyber defenders only, became generally available on Oct 13, 2026. The stable model gemini-4-argon is on the Gemini API paid tier and in AI Studio, in Google Antigravity for AI Ultra subscribers, and in the Gemini app and AI Mode in Search for Pro and Ultra subscribers. Input, thinking and output share one 1,048,576-token window. Synchronous calls default to 65,536 output tokens; with background execution and max_output_tokens above 65,536, the server chains continuation hops of up to 64K tokens each, so a single response can approach 1M tokens. Thinking levels are low, medium (default) and high. Migration notes from Google: raise max_output_tokens to at least 32,768, remove explicit context caching (unsupported; implicit caching discounts cache hits 95%), drop temperature/top_p/top_k (ignored), set client timeouts of at least 15 minutes, and keep multi-turn state with previous_interaction_id. Pricing is $2 input / $10 output per million tokens through Nov 30, 2026, then $4 / $20. There is no free-tier access. All benchmark claims are vendor-reported: #1 on the Vals Index, 87% on ERQA, and 0.7% attack success on Gray Swan's indirect prompt injection benchmark across 15 adaptive attempts. Google's case studies include a 2.85x Gemma 4 E2B speedup on vLLM (A100, ~37,900 to 108,000+ tokens/s) and a 40% cut in ADK long-session prep time. Google is also expanding Fairwind to hundreds more partners and offering 10,000 scientists six months of Antigravity's top usage tier for free. On day one, Cognition added Argon to Devin Desktop and CLI, citing a self-reported 60.5% on FrontierCode 1.1 and 75% off for Devin users through Oct 27.