Google’s new Gemini Pro model has record benchmark scores — again

Google has announced the release of Gemini Pro 3.1, a significant upgrade in its LLM capabilities, surpassing its predecessor in independent benchmark tests, bolstering its position in the competitive AI model landscape against rivals like OpenAI and Anthropic.

Key Points

• Google released Gemini Pro 3.1, a powerful LLM, which is currently in preview and will be widely available soon.
• Gemini 3.1 Pro shows improved performance over its predecessor, Gemini 3, which was also a strong AI tool upon its November release.
• Independent benchmarks, including Humanity’s Last Exam, demonstrate significantly better performance for Gemini 3.1 Pro.
• Brendan Foody, CEO of AI startup Mercor, praised the model, stating it leads the APEX-Agents leaderboard, indicating rapid advancements in AI capabilities for knowledge work.
• The release comes amid increasing competition in the AI model sector, where companies like OpenAI and Anthropic are launching new powerful models.

Relevance

• The competitive landscape of AI continues to evolve, reminiscent of the tech race during the early 2000s.
• The growing emphasis on LLMs reflects broader trends in AI focusing on multi-step reasoning and agentic tasks.
• Recent advancements in AI technology correspond with increasing investments and interest in the field, highlighting the urgency for companies to innovate rapidly.

The launch of Gemini Pro 3.1 positions Google at the forefront of the AI model race, reflecting both technological progress and the ongoing competitive dynamics within the industry.