Yesterday (18th of November) was a big day for AI growth. ✨
For months, everyone was waiting for Gemini 3. The hype was even bigger than for GPT 5.
Marketing efforts
It was clear that tech influencers and AI tool builders had early access and were spreading rumors about how impressive the model was. But with previous LLM launches, the public release often felt disappointing compared to the early access phase. This time, however, the internet was exploding with news about Gemini.
And the first team to capitalize on all that hype was Elon and xAI with the launch of Grok 4.1. The model turned out to be excellent and instantly hit the number one spot on LMArena.
Gemini 3 launch
The dominance did not last long, because the very next day Google dropped the real bombshell. They not only released Gemini 3 (which outperformed Grok 4.1 and took the top spot on LMArena), but also unveiled a brand new agentic coding IDE called Antigravity, a VS Code fork similar to Cursor.
The best part: Google made Antigravity free to use with Gemini 3 Pro (High/Low). You can also choose Claude Sonnet 4.5 or GPT OSS 120B. Having Claude available was extremely useful, because on launch day Gemini 3 was hard to use due to massive traffic. And you still have the small OpenAI OSS model for tasks like adding emojis to logs or fixing typos.
It was a huge day for Google. Both Gemini 3 and Antigravity are genuinely impressive tools and were welcomed very warmly by the tech community.
Conclusions
1. AI is still at the beginning of its development curve.
It took Google only a few months to release a model that clearly outperforms the rest. This kind of rapid progress happens only in the early stages of an industry. It means there is still so much to discover and improve.
2. Usability and execution speed matter the most.
People laughed when Cursor raised billions for a VS Code fork. And now Google - one of the biggest and most influential tech companies - has done the same.
It sends a clear message: it’s all about generating real value and delivering it fast, not reinventing everything from scratch because you believe you can build something better. Imagine how many small companies, bootstrapped or VC funded, would have tried to build their own IDE instead of forking an existing one.
3. Marketing works. Overdelivering works even better.
AI is a hot topic, but it wouldn’t feel this hot without strong marketing.
Even though people love using AI tools, there are now so many LLMs that a new one could easily be ignored. Every model launch needs a well prepared strategy, with buzz generated even before the official announcement.
And this launch day would have been far less exciting if Google had released only the model. We’ve seen many model launches already. But overdelivering with a powerful IDE plus free model access? That’s huge.