Architectural Prompting: the new elements of Gemini 4, the periodic table of AI

|

Emma Potter

The new Gemini 4 family

This is how our new “periodic table” is structured:

  • Gemini 4 Hydrogen: the heir to the Flash band. It is our daily sprinter: fast execution, daily tasks, light coding and immediate responses at very low cost.
  • Gemini 4 Neon (the new one): the heir to the Pro range. The perfect balance. It is designed for complex high-frequency work, in-depth analysis and multi-step reasoning, while maintaining extreme efficiency.
  • Gemini 4 Carbon: brute force (former Ultra band). Reserved for long and complex projects that require maximum effort in reasoning and planning over broad horizons.
  • Gemini 4 Argon: the very first generation of the 4 series, now joined and largely surpassed in efficiency by the new Carbon and Neon.

What really changes with the arrival of Neon?

For those of us who need to manage complex workflows, the arrival of Neon (already available in the subscriber app, on AI Studio and via API) is fantastic news for several reasons:

  • The missing link: until yesterday there was a clear gap between the speed of Hydrogen and the “heaviness” (also in terms of consumption quotas and waiting times) of Carbon. Neon is placed exactly in the middle. It offers absolute frontier intelligence – managing very long contexts of up to 1 million tokens – but consuming about a third less tokens.
  • Four levels of “thinking” (Thinking Levels): now we can literally modulate the “engine rpm” of the car. Neon allows us to decide how deeply it needs to think before answering (low, medium, high by default, up to xhigh for the most difficult analytical, engineering or scientific problems).
  • Working in Tandem (The Orchestrator): In more advanced tools (like Google Antigravity), Neon and Hydrogen start working together. Neon acts as the “director”: it plans, controls and verifies the logic of our project, while delegating the physical and rapid execution of the individual steps to Hydrogen. It’s like having a Senior Architect leading a team of lightning-fast assistants.

The generational leap: Hydrogen against the “old” Pro

To give you a concrete idea of ​​what this leap means, let’s try to make a small practical comparison. Think of the “Pro” models from just a generation ago: exceptional in output quality, but took their time crunching the data. Often, while the machine was analyzing a rough render or structuring a complex reverse prompt, we physically had time to look away from the screen.

Today, Gemini 4 Hydrogen turns this dynamic on its head. It generates responses at an impressive rate (peaking at around 167 tokens per second), but it does so while delivering a logical quality that was once exclusive to the Pro tier. We’re literally getting the acumen, precision, and depth of yesterday’s flagship model, with the instant responsiveness of today’s entry-level model.

And there is a fundamental detail for our daily work: efficiency. Hydrogen consumes very few resources (the famous tokens) to process data. This means we can upload entire project briefs, long chat histories or dozens of material descriptions without fear of exhausting our daily usage quota or crashing the system. It’s like having gone from an old car that guzzles petrol to a latest generation electric sports car: performance and speed have doubled, but the “fuel” required to make it run has been drastically reduced.

In summary

Having tools like this at our disposal changes the way we approach our work. For quick drafts and everyday use where you need instant feedback, Hydrogen remains unsurpassed. But Neon now becomes our new point of reference for structured tasks, the analysis of complex documents and complex workflows, allowing us to have the power of a flagship model without having to wait for the long times of Carbon. Until the next digital evolution!

The weekly column “Architectural Prompting” is edited by experts Luciana Mastrolia, Giovanna Panucci and Andrea Tinazzo
>> If you are interested in these topics, also sign up to the free Linkedin Newsletter AI & Design for Technicians, we’ll talk about it here!