Daily AI Updates: November 18, 2025

Main Content

• Gemini 3 Pro new model release: Ships a sparse Mixture‑of‑Experts (MoE) architecture with a 1M‑token context window and 64K output tokens, a January 2025 knowledge cutoff, stronger multimodal understanding, and leading scores on major benchmarks: SWE‑bench Verified 76.2%, LiveCodeBench Pro Elo 2439, Video‑MMMU 87.6%.

• Gemini 3 Pro pricing updates: Tiered pricing by prompt length — ≤200K tokens: input $2 per million tokens, output $12 per million tokens; >200K tokens: input $4 per million tokens, output $18 per million tokens. Access via the Gemini API, Google AI Studio, and Vertex AI.

• Gemini 3 Pro feature updates: Adds tool use and function calling, delivering an ≈30% improvement relative to Gemini 2.5 Pro. Supports agentic coding and vibe coding. Integrates Google Search, URL context, and a code‑execution tool with composable structured outputs.

• Gemini 3 Pro performance optimization: Achieves HLE 41.0%, GPQA Diamond 93.8%, ARC‑AGI‑2 Verified 45.1%. Improves coding tasks ≈20%, world knowledge ≈15%, screen understanding 6×, and video/image understanding ≈10%.

• Gemini 3 Pro integrations: Adds deep integrations with Cursor, GitHub, JetBrains, Cline, and Google Antigravity, enabling agentic CLI, Android Studio, and bash tooling, strengthening multi‑step workflow automation.

• Gemini 3 Pro industry collaboration: Works with the Google Antigravity IDE for end‑to‑end task planning, coding, testing, and iteration, with support for Nano Banana and Gemini 2.5 Computer Use agent behaviors.

• Gemini 3.0 in Google Colab: Integrates Gemini 3.0 into Google Colab, improving code‑generation accuracy, reasoning depth, and project insights, delivering advanced AI assistance in Python programming environments.