Google has unveiled Gemini 3.1 Pro, its most advanced multimodal reasoning model to date, featuring a massive 1 million token context window and 64K token output capacity. The model represents the next iteration in Google's Gemini 3 series, designed to handle complex tasks across text, audio, images, video, and entire code repositories.

Gemini 3.1 Pro demonstrates significant performance improvements over its predecessor, achieving 44.4% on Humanity's Last Exam for academic reasoning compared to Gemini 3 Pro's 37.5%. The model excels in agentic coding tasks with 80.6% on SWE-Bench Verified and 68.5% on Terminal-Bench 2.0. It also leads in abstract reasoning with 77.1% on ARC-AGI-2 and achieves 94.3% on GPQA Diamond for scientific knowledge.

Available through Google Cloud Vertex AI, Google AI Studio, and Google Antigravity, Gemini 3.1 Pro is optimized for applications requiring advanced coding, long-context understanding, and multimodal reasoning. The model maintains Google's commitment to safety and ethical AI development, with consistent safety performance compared to Gemini 3.0 Pro while reducing unjustified refusals by 0.08%.