Alphabet’s Google has begun rolling out Gemini 4 Argon, its long-awaited flagship artificial intelligence model, but the company is grappling with internal skepticism over how well it performs in key areas, such as coding. Google unveiled the model Wednesday to a small group of trusted cybersecurity partners and said it would expand access after additional testing, starting with paid subscribers. The company said Gemini 4 posted leading scores on several benchmark tests, including beating OpenAI’s Astra model on one that measures security skills.
But some insiders say those metrics don’t tell the whole story. While Gemini 4 has performed well on benchmarks widely used to gauge model efficacy, it does less well when employees actually put it to work, according to people with direct access to the effort. The model struggles to handle certain coding tasks, said the people, who requested anonymity to discuss an internal matter.
Extract — continue reading at the source.
