Google's Gemini 3.1 Pro Doubles Reasoning Performance

Google released Gemini 3.1 Pro, an upgraded artificial intelligence model that doubles reasoning performance and introduces adjustable thinking levels.The new model achieved a 77.1% score on the ARC-AGI-2 benchmark and enables the generation of animated vector graphics from text prompts.Available in preview through Vertex AI and the Gemini app, the update features testimony from Hanlin Tang of Databricks who reported improved results on the OfficeQA benchmark.