This is Gemini 3: our most intelligent model that helps you learn, build and plan anything. It comes with state-of-the-art reasoning capabilities, world-leading multimodal understanding, and enables new agentic coding experiences. 🧵

Our first release is Gemini 3 Pro, which is rolling out globally starting today. It significantly outperforms 2.5 Pro across the board: 🥇 Tops LMArena and WebDev @arena leaderboards 🧠 PhD-level reasoning on Humanity’s Last Exam 📋 Leads long-horizon planning on Vending-Bench 2

🔵 State-of-the-art reasoning It understands and answers prompts with more depth and nuance than ever before. Get clear, direct responses without the clichés and filler. As our most factual model, it’s more reliable for complex questions in science and math.

🔵 World-leading multimodal understanding It can take in any kind of input you give it – from text to video to code – and responds with whatever best suits your needs. Ask Gemini to break down concepts from a long video or quickly turn a research paper into an interactive guide.

🔵 Our best model for vibe and agentic coding Gemini 3 can make beautiful and dynamic apps from a single prompt. It can follow instructions exceptionally well, so it knows exactly what you’re asking for – and how to make it. We also improved agentic code performance to support existing tools – including our new agentic development platform, Google Antigravity.

We’re excited to share Gemini 3 and can’t wait to see what you do with it. Try it now on: 🔘 @GeminiApp 🔘 @GoogleAIStudio 🔘 AI Mode in @Google Search Find out more ↓

What’s the training data?

My Word!!! 😲😲😲 Gemini, the only model I only know that has gottent this correctly

it’s so over

google on their way to dominate the whole market

THERE IT IS ATTAAACK DESTROY IT KILL IT WITH FIRE

We jailbroke it first 😈😈

Fantastic release, well done! 💫

we are so back

Yaay GEMINI 3 is OUTTT

Gemini 3.0 Pro is incredible. I just said "a little character running quickly," and it generated such a beautiful animation.

Summary of Gemini 3 announcement

we really speedran from autocomplete to please build my entire company, thanks.

Wonder how good is it in coding compared to claude and openai

Impressive results already LiveBench will be the real test. Excited to see how Gemini 3 performs on hidden, real-world problems.

we really got Gemini 3 before gta 6

IT’S FINALLY OUT

Deepmind rocks. Elon, sama you're so far behind that you can't even see how they cook. Gemini 3 is so much better that everything I ever tried, it's like a jump to thinking LLMs 2 years ago. It's marvelous. It's scary.

cooked🔥

OH. MY. GOD. You guys cooked!!!

chapeau 🎩

Let's GOOOO!

It's just so good, Gemini 3.0. Love it.

Fire 🔥🔥

It took a while, but it was worth the wait.

Hello Gemini 3.0, we’ve been waiting for you

Gemini 3 = 3-day memory. Here’s what happens when I say farewell to Gemini 2.5 Pro. Marked as “sensitive query” and deleted in 72h.

Nice, my intelligence just got an upgrade

Gemini 3 is finally here!!🔥🔥

Here we go.

Ok, let's start with font design studio

Okay, but what's in the training data?

Yeah! He is the first one who correctly (-2224) solved this problem 👏👏👏 «Find the smallest integer a for which the inequality is not satisfied for exactly 2222 integer values of x»

Show us the training data.