Google's Gemini 4 Argon aims at hard work, from $2 per million tokens

Google has released Gemini 4 Argon, a new top-tier model it says is built for "deep work" in coding, finance, legal, cybersecurity and science. It's open to developers, businesses, scientists and everyday users, so you may soon meet it in tools you already use.

Google's headline claim is that Argon ranks first on the Vals Index, a benchmark covering finance, coding, legal and tax work. The pages we read don't include the scores or rival results, and they don't say who runs the index. We couldn't confirm whether it's an independent third party benchmark or something Google built, so treat it as a claim until outside testers weigh in. Pricing starts at $2 per million input tokens and $10 per million output tokens, with cached input 95% cheaper. The text we read cuts off before saying which tier that applies to, so check Google's pricing page before budgeting.

Most of the evidence comes from Google's own engineers. Inside its Antigravity coding tool, Google says Argon lifted agentic code submissions by more than a third and cut rollbacks by nearly a third. In Chrome, it helped untangle a huge slice of the Chromium codebase, finishing a refactor across hundreds of files in weeks rather than months. On Google's open-source Agent Development Kit, it worked overnight and, Google says, trimmed long-session prep time by 40%.

Google also says it follows style instructions better than earlier Gemini models, hedges less and avoids clichés. Those are subjective claims, and the pages we read offer no side-by-side examples or outside testing to back them up. Demos include a Blender-built pagoda and a racing game made from photos of a real desk, which you can try in Google AI Studio.

What to watch: independent benchmark results and how Argon fares against ChatGPT and Claude outside Google's own repos. We found no independent comparison with those models yet, so whether Argon is actually better is still unanswered. Have you tried it yet, and does it beat what you use now?