Google published its monthly recap of AI announcements on November 2, 2026. It gives the month, not exact dates, for each item. Here are the ones that matter for model watchers, as Google describes them.
New or newly listed
- Gemini 4 Carbon: Google says it targets long-horizon engineering and cybersecurity defense. It is in pre-release testing with the U.S. government and trusted cyber defenders through the Fairwind Program, and Google gives no broader launch date. Update, November 11, 2026: Google made Carbon available to paying developers on November 10 and is rolling it out to Google AI Ultra subscribers; see our summary.
- Genie 4: a real-time interactive world simulation model with native audio, steerable with natural language. Google says demos are on the Google DeepMind blog and states no wider availability.
- EmbeddingGemma 2: an open embedding model with 740 million parameters that maps text, code, images, video and audio into one space on devices.
- Playground: an experimental platform for building and sharing single- and multiplayer games from text prompts.
- Flow Music plugins: lets music producers create custom instruments and audio plugins through vibe coding. Google gives no availability details.
- SynthID Detector: Google says it checks whether an image, video or audio clip was AI-generated, and expanded globally, in English only.
What to know
Several items are preview or research, so do not assume you can use them. We have not seen the individual announcement pages for Carbon, Genie 4 or EmbeddingGemma 2, only Google’s recap, so check Google’s links for details.
Why it matters
It shows where Google is heading after Gemini 4 Argon: a security-focused sibling, world models and on-device embeddings. See the AI model tracker, the Gemini tool page and the related Guided Learning research.
Source: Google. We summarise in our own words; check the source for the full details.