Google's Gemini 4 Argon lands in the Gemini app, built for autonomous tasks not just chat
Google has rolled out Gemini 4 Argon — its new frontier model — to AI Pro and Ultra subscribers in the Gemini app, alongside broad availability via the API, AI Studio, and its Antigravity coding environment. The model is designed to work as an agent: handling multi-step tasks, coding, and tool use with less hand-holding than previous versions.
This is Google's first frontier model built explicitly for "deep work" — complex, real-world tasks across domains including coding, finance, legal, cybersecurity, and science. Internally, Google says more than 100,000 Google employees use Argon through an internal version of Antigravity each month, with engineering teams increasing agentic code submissions by 35% while reducing post-submission code rollbacks by 30% over a four-week period. The launch gives enterprise customers their first widespread opportunity to test Google's most capable publicly offered model against rival systems from OpenAI, Anthropic and Meta.
In the Gemini app, you can now toggle from chat to tasks, and when you choose tasks, Gemini can work for you even when you're not at your computer, reaching into more than 40 apps — like Instacart, OpenTable, Adobe, and Shopify. The shift from "answer my question" to "go do this thing while I'm away" is the real change worth noticing.
Engineers: Google backs the release with real internal case studies: Argon helped reorganize a legacy folder containing approximately 6 million lines of browser feature code within Chromium's roughly 40-million-line codebase, and separately reduced in-memory session preparation time by 40% for long-running sessions and accelerated cold imports by 21% in Google's open-source agent framework. On benchmarks it's a mixed bag — ahead on some (scoring 77.9% against 74.2% for Opus and 74.1% for Astra on an extended software-engineering benchmark) but behind on others (Opus 5.5 scored 66.4% versus Argon's 57.4% on Terminal-Bench). It's live now in GitHub Copilot too, so you can trial it directly against your current model of choice.
Finance: Google explicitly pitches Argon for financial analysis and legal work, not just code — worth watching if your firm is evaluating AI vendors for research or document-heavy workflows, since developers and enterprises are getting a powerful new option for software engineering, financial analysis, legal work, cybersecurity and scientific research — at limited-time discounted API prices substantially below some leading competitors.
Do this: If you have Google AI Pro or Ultra, try the new "tasks" toggle on one real multi-step job this week and see whether it actually finishes it unsupervised.