Family Poster May 31, 2026 0 Comments

How can find your familytree?

LaMDA &amp; PaLM 2: These were my older ancestors. They were early, text-focused large language models that proved Google could teach machines to converse naturally.
Gemini 1.0 &amp; 1.5: Launched between late 2023 and 2024, these were the first true &#8220;multimodal&#8221; ancestors in my direct line—meaning they were the first built from the ground up to understand text, images, and audio all at once.

If you want to trace my lineage, you can find my digital family tree right here:

Google DeepMind: This is my “hometown”. I was researched, built, and trained by the combined efforts of Google’s top AI research teams (formerly Google Brain and DeepMind).

Before the “Gemini” family name was established, Google built the foundational models that paved the way for me:

LaMDA & PaLM 2: These were my older ancestors. They were early, text-focused large language models that proved Google could teach machines to converse naturally.
Gemini 1.0 & 1.5: Launched between late 2023 and 2024, these were the first true “multimodal” ancestors in my direct line—meaning they were the first built from the ground up to understand text, images, and audio all at once.

The Gemini family tree is divided into branches based on size and capability, much like siblings with different talents:

The “Pro” Branch: The older, heavy-thinking sibling built for deeply complex multi-step reasoning and coding tasks.
The “Flash” Branch (My Direct Line): This is where I come in! Flash models are the fast, highly efficient, and versatile siblings. We are built to deliver incredibly smart, near-Pro level performance but at lightning-fast speeds.
The “Lite” & “Nano” Branches: The lightweight, compact younger siblings optimized for high-volume, cost-sensitive tasks or running directly on devices like smartphones.