Google releases Gemini 4 Argon globally as flagship AI model

Google has globally launched its flagship AI model Gemini 4 Argon, making it available to paying subscribers and developers via API. The launch comes after months of delays, internal leadership restructuring at DeepMind, and the cancellation of Gemini 3.5 Pro. Google offers introductory API pricing of $2 per million input tokens through November 30, 2026. While Argon tops professional task leaderboards like the Vals Index, independent benchmarks show mixed results against competitors like Anthropic's Claude Opus 5.5.

Gemini 4 Argon global rollout

• ▪Google's release of Gemini 4 Argon comes 13 days after Google's initial September 30, 2026 announcement, during which access was restricted to cybersecurity defenders through its Fairwind Program
• ▪Google is expanding its Fairwind Program to hundreds of additional partners, granting vetted cybersecurity defenders access to a version of Gemini 4 Argon without cyber guardrails
• ▪Google is offering 10,000 scientists and researchers six months of free access to the highest usage tier of Google's Antigravity agentic development platform on a rolling basis
• ▪Google broadly released its flagship Gemini 4 Argon multimodal AI model globally on October 13, 2026, making it available to developers, businesses, and consumers

API pricing tiers

• ▪Google offers a unified API limit of up to 1 million combined input and output tokens for Gemini 4 Argon using a specialized continuation protocol for demanding tasks
• ▪Google is offering Gemini 4 Argon at introductory API rates of $2 per million input tokens and $10 per million output tokens, with a 95% discount on cached input tokens
• ▪Google confirmed that introductory API pricing for Gemini 4 Argon expires November 30, 2026, doubling on December 1 to $4 per million input tokens and $20 per million output tokens

Subscription access restrictions

• ▪Google restricts Gemini 4 Argon access on Google's Antigravity agentic development platform to Google AI Ultra subscribers, while retaining Gemini 3.1 Pro for other tiers
• ▪Under Google's updated model tier structure for the Gemini app, free Gemini users are restricted to Flash-Lite, while paying Plus subscribers have access only to Flash models
• ▪Access to Gemini 4 Argon in the Gemini app and AI Mode in Search is restricted to paying subscribers on Google AI Pro ($19.99/month) and AI Ultra tiers

Internal Google deployment results

• ▪Google used Gemini 4 Argon to reorganize Chromium's legacy browser code across hundreds of files in its 40-million-line codebase, finishing in weeks a project expected to take months
• ▪Google's Agent Development Kit team used Gemini 4 Argon to optimize the Agent Development Kit's Python framework, reducing in-memory session preparation time by 40% and accelerating cold imports by 21%
• ▪More than 100,000 Google employees use Gemini 4 Argon monthly via an internal Antigravity tool, increasing agentic code submissions by 35% and reducing rollbacks by 30%

Competitive benchmark performance

• ▪Google reported that Gemini 4 Argon achieved an 87% score on Google's Embodied Reasoning Question Answering benchmark and instructed Apptronik's Apollo humanoid robot to land seven consecutive cornhole throws
• ▪Gemini 4 Argon ranks first on the GDP-weighted Vals Index with 68.90%, ahead of Claude Sonnet 5.5 at 67.04% and Claude Opus 5.5 at 66.97% as of October 12, 2026
• ▪Artificial Analysis assigns Gemini 4 Argon a score of 53 on the Artificial Analysis Intelligence Index, tying OpenAI's GPT-6 Astra but placing below Claude Opus 5.5 at 58 and Meta's Muse Wonder Full at 56

DeepMind leadership restructuring

• ▪Google released Gemini 4 Argon following a DeepMind reorganization where co-founder Demis Hassabis became Alphabet chairman and chief scientist, and Koray Kavukcuoglu assumed DeepMind leadership
• ▪Google cancelled its previously promised Gemini 3.5 Pro model prior to releasing Gemini 4 Argon, making Argon its first major frontier release since Gemini 3 in late 2025
• ▪Google employees are internally testing updated versions of Gemini 4 Argon under codenames Barium and Carbon that show enhanced coding capabilities over Argon

Debatable claims

• ▪Distributing guardrail-free AI models to cybersecurity defenders poses unacceptable security risks
• ▪Google should provide free access to flagship AI models