# Google ships Gemini 4 Argon, starting with cyber defenders

> Google DeepMind announced Gemini 4 Argon, its first new frontier model since Gemini 3, on 30 September.

*Google DeepMind's first new frontier model since Gemini 3 reaches Fairwind Program members first, with paid API and AI Ultra access to follow.*

By Behzad Hosseini · FeaturedDaily
Canonical: https://featureddaily.com/news/google-ships-gemini-4-argon-starting-with-cyber-defenders

**What happened:** On 30 September 2026, Google DeepMind [announced Gemini 4 Argon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/). It is the company's first new frontier model generation since Gemini 3. Google is aiming it at long-horizon coding, legal and finance work, and cyber defence.

**The rollout:** Trusted cyber defenders in Google's Fairwind Program get it first, along with US government pre-release access. Paid API users and Google AI Ultra subscribers follow. Google says broader developer, enterprise and consumer release is "coming soon".

**The numbers:** Introductory pricing is $2 per million input tokens and $10 per million output tokens, with cached input 95% off. After the introductory period, it rises to $4 and $20. The output limit goes to 1 million tokens, up from 64K, which Google bills as industry-leading.

**The benchmarks:** These are Google's own figures.

- DeepSWE v1.1: 77.9%
- CWE-bench v1 (vulnerability remediation): 68%, tied first
- AutomationBench: first, at 51.3%
- LVBench (long video): 91.7%

Google also says Argon leads the Vals Index across finance, coding, legal and tax, and leads Gray Swan prompt-injection robustness.

**The rivals:** Google positions Argon against OpenAI's GPT-6 Astra and Anthropic's Claude Opus 5.5.

**In their words:** Koray Kavukcuoglu, SVP of Google DeepMind and Chief AI Architect, wrote in the launch post: "Argon is helping our quantum computing researchers optimize the spacetime resources (qubits × gates)"

**The catch:** Bloomberg reported that some Google employees privately doubt Argon's real-world coding ability. Google has not confirmed this. [CNBC](https://www.cnbc.com/2026/10/02/tech-download-google-argon-frontier-openai-anthropic.html) asked whether Argon can catch the frontier. One analyst said "the final proof will be in enterprise production environments once the model is fully released".

**What's next:** Google has given no date for the wider release to developers, enterprises and consumers. The introductory prices apply until the higher rates start.

## Key takeaways

- Argon reaches cyber defenders first, then paid API and Ultra users
- Introductory pricing: $2 input, $10 output per million tokens
- Output limit rises to 1 million tokens from 64K

## Sources

- [Gemini 4 Argon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-argon/) — Google, 2026-09-30
- [Can Google's new model really catch up to OpenAI and Anthropic at the frontier?](https://www.cnbc.com/2026/10/02/tech-download-google-argon-frontier-openai-anthropic.html) — CNBC, 2026-10-02