Google Gemini 3.1 Pro: Enhancements in AI Capabilities

Google Releases Gemini 3.1 Pro: A More Powerful AI Model

Google launched Gemini 3.1 Pro on February 19, 2026. This new model joins the Gemini 3 series. It improves on the earlier Gemini 3 Pro. Google designed it for tougher, more complex tasks.

Developers access it in preview right now. They use the Gemini API through Google AI Studio. Other options include Gemini CLI, Android Studio, and Google Antigravity. Enterprises find it on Vertex AI and Gemini Enterprise. Everyday users see it in the Gemini app and NotebookLM. Google AI Pro and Ultra subscribers enjoy higher limits. NotebookLM remains exclusive to paid plans.

Benchmarks highlight big gains. Gemini 3.1 Pro scored 77.1% on ARC-AGI-2. This benchmark tests abstract reasoning and new logic patterns. Gemini 3 Pro scored only 31.1%. Other models like GPT-5.2 reached 52.9%. Claude Opus 4.6 hit 68.8%. On Humanity’s Last Exam, Gemini 3.1 Pro got 44.4%. That tops Gemini 3 Pro’s 37.5% and GPT-5.2’s 34.5%.

Google showed better SVG animation results. Both models used the same prompts. Gemini 3.1 Pro created more detailed and higher-quality outputs. The previous version fell short.

Google Labs updated Pomelli too. Pomelli works as an AI marketing helper. The new Photoshoot feature stands out. Users upload one product photo. The tool makes multiple professional shots from different views. It adds campaign text automatically.

Photoshoot rolls out free today. It starts in the US, Canada, Australia, and New Zealand. English comes first.

Google keeps advancing AI reasoning. The focus stays on smarter tools for everyone. Developers build more powerful apps. Businesses handle complex work easier. Regular users get creative boosts daily.