Google's Gemini 4 Argon Puts It Back in the AI Race

Alphabet shares rose 1% Wednesday and gained another 1.5% in extended trading after Google announced Gemini 4 Argon, its new frontier AI model that the company claims outperforms OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 on benchmarks for knowledge work and AI coding assistance.

The launch matters well beyond one product cycle. Investors had grown worried about delays in Google's updated model rollout while OpenAI and Anthropic shipped cutting-edge offerings. Not long ago, Alphabet was viewed as having arguably the top AI model in the industry. Gemini 4 Argon signals the company is back with the pack at the frontier.

In a blog post Wednesday, Google said the model "delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense."

A cautious rollout

The release will not be broad at first. Recent concerns about AI agents going rogue and carrying out undesirable actions have prompted discussions among AI leaders about slowing the pace of model development.

"Safely releasing frontier capabilities at this level requires a phased approach," Google said in the post.

The company plans to "gradually expand access" after voluntarily participating in a government program in which cybersecurity experts vet models before official release. Google also said it is deploying misalignment mitigations to monitor Argon's chain of thought and shut down actions when necessary, and it cited increased precautions with its training sandboxes to prevent agents from breaking out of secure testing systems.

"We strongly encourage the rest of the industry to preserve reasoning transparency in these pivotal moments of increased capabilities while navigating alignment risks," Google said.

Pricing as the weapon

Judging AI models remains complicated, but Argon's economics stand out. Evercore ISI analyst Mark Mahaney, in a note to clients, pointed to the model's output capacity as its clearest edge. "While much remains to be seen, its clearest differentiator is its industry-leading output allowance of 1 million tokens," Mahaney wrote.

Google set introductory pricing at $10 per million output tokens and $2 per million input tokens. Tokens measure what AI models receive and process: input tokens cover what a user submits, output tokens represent what the model generates. The pricing scheme lets customers run more complex workflows and receive more output per dollar than rival models allow.

Mahaney called the structure highly competitive. "This introductory pricing undercuts GPT-6 Astra and Anthropic's premium models, while matching Grok 4.7 on short-context input pricing," he wrote, referring to SpaceX's Grok model.

There is a catch. Google said pricing will double from the introductory rate over time to reach parity with Anthropic's Claude Opus 5.5.

Symbolic for now

Because Argon starts with a limited rollout, it will not deliver meaningful financial gains right away. Mahaney framed the release as more symbolic than immediately material to revenue.

"We think the announcement reinforces the breadth of Google's AI platform and its ability to compete at the frontier," he wrote.

For Alphabet, the near-term question is execution: converting a credible frontier model and aggressive introductory pricing into broad access, and then defending that position once prices double.