Google’s Gemini 3.1 Pro Preview Ranks First in AI Performance Index at Under Half the Cost of Competitors

Google’s Gemini 3.1 Pro Preview has emerged as a frontrunner in the Artificial Analysis Intelligence Index, scoring 57 points, four points ahead of Anthropic’s Claude Opus 4.6 and six ahead of GPT-5.2, all while being cost-effective at $892. This model excels in six out of ten categories, including agent-based coding and scientific reasoning, with a notable 38-point reduction in hallucination rates compared to its predecessor, Gemini 3 Pro. Although it requires only 57 million tokens, significantly less than its competitors, it still lags in real-world agent tasks. Internal tests reveal that Gemini 3.1 Pro struggles with fact-checking, verifying only about 25% of statements, lower than both Claude Opus 4.6 and GPT-5.2. Thus, while benchmarks provide insights, users should conduct their own evaluations for comprehensive assessments. Stay informed with the latest AI developments through our exclusive content by subscribing to THE DECODER.

Source link