On Thursday, Google debuted the latest iteration of Gemini Pro, its advanced large language model (LLM). The 3.1 version is now available as a preview and will soon be publicly released, according to the company.
The new model from Google might be among the most powerful LLMs so far. Observers have remarked that Gemini 3.1 Pro represents a significant improvement over its predecessor, Gemini 3, which was already viewed as a highly proficient AI tool upon its release in November.
On Thursday, Google also unveiled data from independent benchmarks, such as one named Humanity’s Last Exam, indicating its superior performance compared to the previous version.
Gemini 3.1 Pro received commendation from Brendan Foody, CEO of AI startup Mercor, whose benchmarking platform, APEX, is intended to assess how effectively new AI models execute real professional tasks. “Gemini 3.1 Pro now leads the APEX-Agents leaderboard,” Foody stated in a social media post, noting that the model’s remarkable results demonstrate “how rapidly agents are advancing in real knowledge work.”
This release occurs as the competition in AI models intensifies, with tech firms continuing to introduce increasingly powerful LLMs intended for agentic work and multi-step reasoning. Other leading companies, including OpenAI and Anthropic, have also recently launched new models.
Techcrunch event
Boston, MA
|
June 9, 2026