Does Google’s new model catch up to OpenAI, Anthropic at the frontier?
This report is from this week’s The Tech Download newsletter. Like what you see? You can subscribe here.
2026 has been a turbulent year for Google’s AI hopes and dreams.
Things started well. Gemini 3, which launched towards the end of 2025, put Google very much in the frontier AI conversation.
But for much of the rest of this year, it’s Anthropic and OpenAI that have been at the forefront of AI advances.
Google will undoubtedly be hoping the release of a new flagship model, Gemini 4 Argon, changes that.
Samuel Boivin | Nurphoto | Getty Images
The frontier
When Demis Hassabis stepped down as CEO of DeepMind, Google’s AI lab, in August, I wrote that new boss Koray Kavukcuoglu was inheriting a race to catch-up with Sam Altman and Dario Amodei’s companies.
Now Google has touted the model as performing better than top OpenAI and Anthropic offerings in some benchmarks.
Artificial Analysis Intelligence Index, a composite benchmark score, places Gemini 4 as lagging behind only Claude Opus 5.5 and Claude Sonnet 5.5 on its leaderboard.
“Google Gemini 4 Argon shows advanced reasoning on critical tasks, according to benchmarks, particularly legal reasoning, finance and other aspects of enterprise knowledge work, including long-running tasks,” Tim Law, director of research for AI at IDC, told me.
Lian Jye Su, chief analyst at Omdia, said that while Google was “late to the cybersecurity and coding game,” Gemini 4 took the company to the frontier in AI, particularly in cybersecurity.
Cybersecurity
Google has been keen to talk up that point.
The company plans to launch the new model in phases, starting with trusted cybersecurity partners while working with the U.S. government on pre-release safety evaluations.
“Starting this rollout in this way gives us more confidence, but also enables us to put a model that is trained and strong in cyber defense in the hands of defenders as soon as possible,” Tulsee Doshi, Google’s Gemini model product lead, told CNBC.
The numerous security incidents caused by rogue AI systems made by OpenAI and Anthropic have created an opportunity for Google to “stake its claims as the trusted model provider for safe and secure AI operations,” Su said.
But while Argon has made Google competitive again in AI, said Nick Patience, AI lead at the Futurum Group, the model doesn’t make it the “leader.”
“At a minimum, though, Gemini 4 Argon opens up another competitive front for enterprise knowledge work with other leading frontier models,” said Law.
Google said in its release that its new model is already being used internally to optimize memory at Google’s data centers, freeing up hundreds of terabytes of memory without buying additional hardware. Quantum computing researchers have also utilized the model.
But the real test, as it always is with new models, is when it sees the cold light of day and is launched more widely. The company has given no indication when that could be.
“While Google touts the success of internal use cases, the final…
Read More: Does Google’s new model catch up to OpenAI, Anthropic at the frontier?