Google Launches Gemini 3.1 Pro with Enhanced Reasoning and Coding Capabilities

February 19, 2026

Details

  • Google AI Developers announced the rollout of Gemini 3.1 Pro on February 19, 2026, featuring a massive boost in intelligence for coding challenges and a new medium thinking level to balance reasoning with latency.
  • The model generates website-ready, animated SVGs from text prompts using pure code, ensuring crisp scalability and small file sizes.
  • Available now in preview via Gemini API in Google AI Studio, Vertex AI, Gemini CLI, Android Studio, and Google Antigravity agentic platform.
  • Benchmarks show 77.1% on ARC-AGI-2 (double Gemini 3 Pro), 94.3% on GPQA Diamond, 80.6% on SWE-Bench Verified, and 2887 Elo on LiveCodeBench Pro.
  • Supports 1M-token context window for text, images, video, audio, PDFs, and code; optimized for agentic workflows, software engineering, finance, and spreadsheets.
  • Enterprises access via Vertex AI and Gemini Enterprise; consumers through Gemini app and NotebookLM, with higher limits for Pro/Ultra plans.
  • Also rolling out to GitHub Copilot for Pro, Pro+, Business, and Enterprise users in VS Code, Visual Studio, github.com, and mobile apps.

Impact

Google's Gemini 3.1 Pro rollout pressures rivals like OpenAI and Anthropic by delivering benchmark-leading scores such as 77.1% on ARC-AGI-2—more than double its predecessor—and strong agentic coding performance comparable to GPT-5.2 and Claude Sonnet 4.6/Opus 4.6, narrowing the gap in reasoning and tool use amid OpenAI's ChatGPT updates and Anthropic's headline-grabbing releases. The introduction of a medium thinking level and 1M-token context optimizes for real-world developer workflows, lowering latency while boosting efficiency in software engineering, finance modeling, and multimodal tasks, potentially accelerating adoption in enterprise platforms like Vertex AI and GitHub Copilot. This positions Google to capture more market share in agentic AI systems, influencing R&D toward reliable long-horizon planning and on-device inference over the next 12-24 months, especially as preview feedback shapes general availability.

Read more