Summary
Google’s Gemini 4 Argon has arrived, and Paul J Lipsky shares what he found after more than a week of early access. Google positions Argon as its most powerful frontier model, built for long, complex work in coding, finance, legal, cybersecurity, and creative writing. According to Google’s published results, it leads on 13 of 19 benchmarks and ties for first on one more, with knowledge work and long context as standout strengths.
The video focuses on real-world testing. Using Google’s Antigravity coding app, Lipsky builds a top-down paintball capture-the-flag game, a Star Fox-style space shooter, and a game controlled by brain-wave data from a meditation headband over Bluetooth. He notes that Argon needs more detailed prompting than Opus 5.5 to produce polished visuals. He also has Argon audit an app built with Opus 5.5, where it finds several bugs, including a potential webhook exposure.
Beyond coding, the video covers Argon’s rollout to Google AI Pro and Ultra subscribers in the Gemini app, where it sounds more natural, stays more grounded in supplied documents, and persists longer on tasks. It also looks at how Argon improves Gemini Spark, including research and PowerPoint generation.
📺 Source: Paul J Lipsky · Published October 13, 2026
🏷️ Format: Review
