Overview
- Google launched Gemini 3 on Nov. 18 and rolled it into Search, the Gemini app, Vertex AI, AI Studio and enterprise offerings.
- The company reports top scores on LM Arena, Humanity’s Last Exam and GPQA Diamond, while experts urge caution on public benchmarks and the higher‑reasoning Deep Think mode remains in safety review before a broader release to AI Ultra subscribers.
- Early developer feedback and independent demos highlight multimodal strengths in coding, richer web UIs, 2D‑to‑3D voxel asset creation, interactive simulations and video analysis such as golf swing breakdowns.
- OpenAI’s CEO told staff to expect “temporary economic headwinds” as CFO Sarah Friar cited a recent dip in ChatGPT engagement, with the company still targeting $13 billion in 2025 revenue alongside an estimated $8.5 billion annual burn.
- Analysts cite Google’s full‑stack position across research, custom chips, cloud and consumer distribution as a scaling advantage over rivals that depend on partners.