Google launches Gemini 4 Argon, its most advanced AI model yet
What happened
Alphabet unveiled Gemini 4 Argon on Wednesday, its most advanced AI model to date and the first of the Gemini 4 generation, focused on real-world software engineering, enterprise knowledge work (finance, legal) and cybersecurity defense. Google claims Argon sets a new record on a real-world software-engineering benchmark, ties for first on cybersecurity tests, and beats OpenAI's GPT-6 Astra and Anthropic's Fable 5.1 on the Vals Index of professional-work performance. The standout spec is a 1-million-token output limit (up from 64K on earlier Gemini models — roughly 16x larger context), with introductory API pricing of $2 per million input tokens and $10 per million output tokens (rising to $4/$20 after the introductory period). Rollout is phased: trusted cyber defenders first via Google's Fairwind Program, with paid API customers and Google AI Ultra subscribers later, as Google completes pre-release safety evaluation with the U.S. government. GOOGL ticked slightly higher in after-hours trading on the announcement.
Why it matters & what’s next are part of Pro, along with the audio summary.
Start your free 14-day trialWho’s affected
- GOOGL (Alphabet) — direct beneficiary; frontier-model benchmark leadership supports its enterprise-AI and Google Cloud monetization story. OpenAI (private) and Anthropic (private) — competitive pressure on their flagship models GPT-6 Astra and Fable 5.1. AI-infrastructure complex (NVDA, AMD, AVGO, MSFT, AMZN, META, ORCL) — sustained frontier competition reinforces the AI capex buildout narrative. Cybersecurity partners/software — Argon is initially being seeded to defensive cyber partners.