Gemini 4 Argon Launches at Half the Price of Claude Opus 5.5

Google announced its newest AI model, Gemini 4 Argon, on 30 September 2026, claiming the lead on 13 of 18 published benchmarks against Anthropic's Claude Opus 5.5 and OpenAI's GPT-6 Astra. For businesses already using AI tools, the model matters less for its ranking and more for its price: Google is launching it at roughly half of what Anthropic charges for its top model.

What did Google announce with Gemini 4 Argon?

Gemini 4 Argon is built, in Google's own words, for deep reasoning across complex, long-horizon workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defence. The headline technical change is output length: Argon can write up to 1 million tokens in a single answer, 15 times the 64,000-token limit on earlier Gemini models, which Google says lets the model finish a full code migration or a long legal document in one pass instead of splitting the job into chunks.

Built to sustain deep reasoning across complex, long-horizon workflows, Argon is fundamentally changing the way we work and build at Google. It delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.from Koray Kavukcuoglu, SVP, Google DeepMind and Chief AI Architect, Google (Google, 30 September 2026)

What can Gemini 4 Argon actually do?

Google's own benchmark table, independently reproduced by VentureBeat, shows Argon leading outright on 12 of 18 published tests and tying for first on one more. Its biggest margins are in business and office work rather than raw programming: on Harvey's Legal Agent Benchmark it scores 19.6 percent against 3.8 percent for Claude Opus 5.5, and on Zapier's AutomationBench, which measures end-to-end business automation, it scores 51.3 percent against 42.5 percent for Opus 5.5 and 41.4 percent for GPT-6 Astra. On real-world long-horizon software engineering (DeepSWE v1.1) it scores 77.9 percent, ahead of both rivals, though GPT-6 Astra still leads on two narrower coding benchmarks and Claude Opus 5.5 still leads on terminal-based coding tasks.

• Cybersecurity defence: Argon ties for first place on CWE-bench v1, which tests finding and patching software vulnerabilities, scoring 68 percent.
• Financial and legal research: it leads Vals Finance Agent v2 at 65.4 percent and the Harvey Legal Agent Benchmark by a wide margin.
• Long documents and video: it leads GraphWalks (reasoning over up to 1 million tokens of text) and LVBench (long video understanding).

Does this matter for a small business?

Not immediately, because access is restricted. Google is rolling Argon out first to trusted cybersecurity defenders through its Fairwind Program, launched 2 September 2026 with more than 650 partner organisations, while it takes part in the United States government's voluntary pre-release testing process. Paid API customers and Google AI Ultra subscribers come next, with no date announced. Cybersecurity firm Wiz, one of the first named Fairwind partners, has already used Argon through its Scan for Good initiative to uncover a critical vulnerability exposing sensitive personal information in hospital software used worldwide, a risk that earlier frontier models had missed. That is a preview of what any business should expect once broader access opens: a cheaper, longer-reasoning model aimed squarely at the kind of coding, admin and compliance work most businesses already pay AI tools to do.

Also today

• OpenAI is reportedly in talks to raise 30 billion US dollars in private funding at a 1.4 trillion US dollar valuation as its IPO plans stall, according to a Bloomberg report (BPDATA AI News).
• US President Donald Trump signed a voluntary White House AI accord on 30 September 2026 encouraging AI companies to self-police on cybersecurity and biosecurity safeguards, with no binding enforcement (Al Jazeera).

What to do this week

You cannot buy access to Gemini 4 Argon yet, so there is nothing to switch to today. What is worth doing is watching the Gemini API pricing page over the next few months: if Argon's launch price holds once it reaches paid customers, every provider your business already uses, including Claude and ChatGPT, will likely come under pressure to match it. The bigger lesson from this launch is that the tools worth building your workflows around today, like Claude Projects, Cowork and Skills, are the same category of tool Argon is chasing: AI that can hold a long, complex task from start to finish without constant supervision. In our Master Claude AI for Business Productivity program, we show Sabah business owners how to put that kind of AI to work on real admin, finance and reporting tasks now, so switching tools later is a pricing decision, not a rebuild.