Google is rolling out its powerful Gemini 4 Argon AI model to the public globally. You can now use it on your phone via the Gemini app. Josh Woodward said Argon “answers questions more directly with less hedging.” It handles complex coding and cybersecurity tasks while outperforming many rival models.
Google is rolling out its most powerful AI mode, Gemini 4 Argon, to the public. The company says Argon is one of the most advanced models out there, competing against the likes of Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6 Astra. This comes nearly two weeks after Google first announced Gemini 4 Argon.
The company says Gemini 4 Argon is built for “deep work in complex, real-world tasks” across different domains, such as coding and cybersecurity. As per Google, Argon takes the number 1 spot on the Vals Index, which tracks model performance across sectors such as finance, coding, legal and tax work.
You can now use Gemini 4 Argon
Gemini AI Pro and Ultra users can now use Gemini 4 Argon via the Gemini app and AI Mode in Google Search. On the Gemini app, you can connect the AI to over 40 apps, including Adobe and Shopify. The AI can then do tasks for you, such as ordering groceries and booking tables. This is similar to how the Meta’s Muse AI works.
Josh Woodward, Vice-President of Product for Gemini and Google Labs, said Argon “answers questions more directly with less hedging, doesn’t repeat itself more than necessary, and skips clichs throughout the responses.” Gemini 4 Argon also gets stronger coding features and tighter links to Google Antigravity.
For developers, the model is available through the Gemini API, and to Google AI Ultra subscribers through Google Antigravity. Argon comes at an introductory price of $2 (roughly Rs 190) per million input tokens and $10 (roughly Rs 950) per million output tokens, with cached input tokens offered at a 95 per cent discount. After November 30, 2026, the company said the price will move to $4 (roughly Rs 380) per million input tokens and $20 (roughly Rs 1,900) per million output tokens for the public-access version of the model.
Google has also teased Gemini 4 Argon of being capable of designing 3D Android UIs, including “live frosted glass shaders to dynamic lighting and ripple effects.”
How good is Gemini 4 Argon?
In benchmarks like AutomationBench and coding test DeepSWE 1.1, Gemini 4 Argon can outperform GPT-6 Astra and Claude Opus 5.5. Argon also gets an industry-leading 1 million token output limit, compared to roughly 1,28,000 that rival AI models get, that allows the AI to give longer responses.
Google says its teams and researchers have already been using Argon. The company claims software engineers using Argon in Antigravity increased agentic code submissions by more than 35 per cent while reducing code rollbacks by nearly a third. Gemini 4 Argon also helped the Chrome team separate core webpage handling from a roughly 6 million-line legacy folder within Chromium’s 40 million-line codebase – completing work that would usually take months in a matter of weeks.
Gemini 4 Argon also brings major gains in physical understanding, and visual-semantic reasoning. Google says this can unlock new workflows across the robotics stack and accelerate progress toward physical AGI. On Embodied Reasoning (ERQA) benchmark, which tests a model's ability to understand and solve physical, spatial problems in 3D environments. Argon scores 87 per cent, higher than every other frontier model.
Google also pointed to cybersecurity and science use cases. It said the Fairwind programme is being expanded to hundreds of additional partners. This comes after Google-owned cybersecurity firm, Wiz, used Argon to chain vulnerabilities and fix a flaw affecting healthcare software, helping protect hundreds of thousands of private patient records. Argon was previously limited to select partners as part of Fairwind.
In research, the company said Argon nearly doubled quantum error-correction throughput on AlphaQubit and helped discover a broader class of counterexamples to Grad’s conjecture with Professor Matt Landreman of the University of Maryland. The company also said the model helped design five novel serine esterase enzymes, with laboratory testing confirming catalytic activity across those designs.
Gemini 4 Argon is Google’s first frontier AI model since Gemini 3 Pro which was released last year. The company reportedly cancelled the release of Gemini 3.5 Pro after it failed to meet internal goals.