0:03 Gemini has been multimodal since the
0:06 beginning. Our Gemini 1 models could
0:10 see, hear, understand vast amounts of
0:12 information and natively generate across
0:15 modalities. Gemini 2 took the next
0:17 [music] step with advanced reasoning,
0:20 enabling AI agents to think, code,
0:23 [music] and take action. We're taking
0:24 all of this to the next level with
0:28 Gemini 3. Gemini 3 is the strongest
0:30 model in the world for multimodality and
0:32 reasoning. [music] It's our most
0:33 intelligent model that helps you bring
0:37 any idea to life. Starting today, we're
0:39 rolling out Gemini 3 in our products so
0:42 you can experience it for yourself. In
0:44 Google search, Gemini 3 enables new
0:47 kinds of generative user interfaces. It
0:49 codes interactive simulations like this
0:53 one, customuilt for your search.
0:55 In the Gemini app, you can use the vast
0:57 [music] range of Gemini 3 capabilities
0:59 for all kinds of tasks. You could
1:03 supercharge how you learn, create, plan,
1:06 [music] take action, analyze complex
1:09 videos, and more. You could start
1:11 [music] building with Gemini 3 Pro,
1:14 available in the API today. We're even
1:17 introducing a new platform, Google Anti-gravity.
1:18 Anti-gravity.
1:20 It's our vision of software development
1:23 at the frontier of model intelligence.
1:25 It lets you use Gemini 3's agentic
1:27 coding capabilities to accelerate how
1:30 you build. The rollout of our Gemini 3
1:32 series will continue [music] in coming
1:34 weeks. We'll be introducing models with
1:37 new features across different modalities
1:40 and more model sizes.
1:42 This is just the beginning of our Gemini
1:43 3 series. >> [music]
1:43 >> [music]
1:45 >> It's another critical step on the road
1:48 towards making AI more intelligent and
1:51 personal for you. Learn more about how
1:52 you can try Gemini 3 at deepmind.google/gemini.
