Google just released its flagship AI model, Argon, which is part of the company's latest Gemini 4 series. It comes with a lot of delays after months. The objective here is straightforward: Catch up with Anthropic and OpenAI in the AI race. What we know and what we don't.
What Is Google Argon?
Argon is a new top-tier artificial intelligence model, announced on Wednesday by Alphabet's Google. It is bigger than Google's earlier “Pro” series of sophisticated models, said a spokeswoman for the company. Google dubs it the "most performant model to date" designed for "complex workloads". The concept is simple, just a larger model for more challenging jobs.
Argon vs OpenAI Astra and Anthropic Opus
According to Google, Argon is similar to frontier models such as OpenAI's Astra and Anthropic's Opus in terms of its coding and cyber capabilities. The company beat several industry benchmarks with its scores in its press release.
It's a mixed picture overall. But argon lagged behind in some other measures, such as two of the four coding standards Google selected for its inclusion. And these results are self-reported. It will be tested independently to see how it performs in the real world, away from Google's own labs.
Gemini 4 Release Date: Still Unclear
Google hasn't provided a release date. Until now, Argon is going to choose cybersecurity partners. Google is also participating in the Trump administration's opt-in process for access to pre-release models.
To developers and the general public, that translates to a “when is the Gemini 4 release date” question.
Gemini 3.5 Pro Is Off the Table
Google is no longer going to release Gemini 3.5 Pro. The CEO Sundar Pichai had earlier announced that it would be available in June. This plan has been superseded by a new one, Gemini 4.
DeepMind Leadership Changes
As the day approached, Google revamped its DeepMind AI lab.As time dragged on, Google restructured its DeepMind AI lab. Founder and CEO Demis Hassabis was removed and multiple members of the Gemini model team have been fired.
The changes are not just at the level of a roadmap. These impact the individual(s) performing the work, and can influence the speed at which a team ships.
Google Gemini vs Anthropic and OpenAI: The Cost Angle
Anthropic and OpenAI continued to plow the research frontier and release new versions of their best models during the time Google spent on delays. In July, Pichai challenged that notion that Google was falling behind in the AI race.
Nevertheless, there has been a change in Google's messaging in recent months. It does not focus so much on Gemini's high-tech features but rather on the cost competitiveness with rival models.
Cost matters. Businesses with significant AI workloads keep an eye on the cost per task. If it's not top on all the benchmarks, a less expensive model that gets the job done may end up winning the business.
What to Watch Next
- Google has not provided any timetable for Argon.
- Independent benchmarks: Google's self-reported scores will be tested using third-party results.
- The price: It is now Google's business card.
- Feedback from cybersecurity partner users: Real-world strengths and gaps can be identified by early users.
- DeepMind stability: A new team of leaders will need to get back to its roots.
The Bottom Line
The results of Argon demonstrate Google's continued dominance in the highly competitive field of AI. It is a larger machine designed for complex workloads and has excellent coding and cyber benchmark claims. However, it's late, it's not publicly released, and it's not always the best comparison. The coming months will tell if the late announcement can build momentum for Gemini 4, or if a company with a transition can keep up with quickly-moving competitors.
