On February 19, Google launched Gemini 3.1 Pro, enhancing its AI model’s reasoning capabilities by over 200% while maintaining pricing. Notably, it achieved a score of 77.1% on the ARC-AGI-2 benchmark, a significant leap from Gemini 3 Pro’s 31.1%. This model is now accessible on Google’s platforms, with elevated usage limits for AI Pro and Ultra users. Priced at $2 per million input tokens and $12 per million output tokens, it remains competitive against OpenAI and Anthropic offerings.
Gemini 3.1 Pro excels on 12 of 18 benchmarks, including 94.3% on GPQA Diamond and 2,887 Elo on LiveCodeBench Pro. Its dynamic thinking feature adjusts reasoning depth based on prompt complexity, enhancing efficiency. Despite slight advantages from competitors, Google leverages its vast user base through integrated services like Gmail and Docs. The model’s performance in real-world applications will ultimately determine its impact, as its high benchmark scores suggest a strong capability in handling novel tasks.
Source link
