This past Thursday, Google unveiled the latest iteration of Gemini Pro, its formidable large language model. Version 3.1 of this model is presently accessible for early access and, as stated by the corporation, is slated for a full public launch shortly.
This recent development from Google could prove to be one of the most formidable large language models to date. Observers have pointed out that Gemini 3.1 Pro seems to represent a significant advancement over its forerunner, Gemini 3—a model which, when it debuted in November, had already been deemed an exceptionally proficient artificial intelligence instrument.
Moreover, this past Thursday, Google additionally presented figures from unbiased evaluations—including an assessment named Humanity’s Last Exam—which demonstrated its substantially enhanced performance compared to its prior edition.
Gemini 3.1 Pro also garnered acclaim from Brendan Foody, the chief executive of the artificial intelligence nascent company Mercor, whose evaluation framework, APEX, is configured to assess the proficiency of novel AI models in executing genuine professional duties. “Gemini 3.1 Pro has now ascended to the pinnacle of the APEX-Agents ranking,” Foody declared via a social media update, further noting that the model’s remarkable outcomes illustrate “the rapid progress agents are making in authentic intellectual labor.”
This unveiling transpires amidst an intensifying competition among AI models, with technology firms persisting in launching progressively potent large language models tailored for autonomous operations and complex logical deduction. Further key players—including OpenAI and Anthropic—have likewise introduced fresh models lately.
Techcrunch event
Boston, MA
|
June 9, 2026
{content}
Source:{feed_title}