Skip to content
Advertisement
Gemini 4 Neon surfaces in Google Cloud billing and developer tools ahead of launch

Gemini 4 Neon surfaces in Google Cloud billing and developer tools ahead of launch

gemini icon

Google's new mid-tier Gemini 4 model has a price tag and a command-line slot but no official announcement yet

Google has a new Gemini model waiting in the wings. It showed up the way unreleased products often do: in the plumbing.

Gemini 4 Neon has been added to the Antigravity CLI as a Pro orchestrator model. It has also appeared in Google Cloud’s public SKU and billing catalog, so the price is public even though Google hasn’t formally announced the model.

What the billing catalog reveals

Neon landed in Google Cloud’s pricing catalog on November 18, 2026. It is listed at $3 per million text input tokens and $15 per million text output tokens.

A quick primer: tokens are the small chunks of text that AI models read and write. Developers pay for what goes in and what comes out, so those two numbers set the cost of every request.

Neon’s rates are exactly double those of Gemini 4 Hydrogen. That model was released on November 12, 2026, at $1.50 for input and $7.50 for output per million tokens.

Neon is anticipated to serve as a balanced mid-tier option. It would sit between Hydrogen, the cost-efficient pick, and the more capable earlier models, Argon and Carbon.

Advertisement

The pricing gap above Neon is about to widen. Argon and Carbon are scheduled to move to standard pricing of $4 per million input tokens and $20 per million output tokens on December 1, 2026.

Once that change takes effect, Neon would sit at a 25% discount to those two models on both input and output.

An orchestrator in the command line

The Antigravity CLI integration offers the clearest hint about Neon’s intended job. In version 1.5.11 of the tool, the model appears as “gemini-pro-orchestrator-neon,” designated for orchestration workflows.

A CLI, or command-line interface, is a text-based tool developers use to run software straight from a terminal. An orchestrator model generally refers to the model that directs a multi-step workflow and decides what happens next.

What’s still unknown is most of what a developer would use to compare models. Google has not released an official announcement or a model card for Neon.

Its context window, multimodal support and benchmark results remain undisclosed. The context window is how much information a model can work with at once. Multimodal support covers whether it handles inputs beyond text, like images.

Google’s periodic table of models

The Gemini 4 lineup follows a naming convention built around noble gases and other elements. Hydrogen, Neon, Argon and Carbon all fit the theme.

Google uses those names to signal different capabilities and price tiers within the family.

What this means for developers and the AI race

For developers, the immediate story is choice. With Hydrogen at the low end, Neon in the middle and Argon and Carbon at the top, teams can match the model to both the task and the budget.

The December 1 repricing is the date to circle. Teams running workloads on Argon or Carbon will see those models move to the $4 and $20 standard rates, and Neon’s 25% discount gives them an obvious alternative to test.

That math only works if Neon performs. Without public benchmarks, nobody outside Google can say how close it gets to the pricier tiers, or where it falls short.

The orchestrator label adds another wrinkle. If Neon is built to coordinate multi-step workflows, developers may weigh it on how well it manages tasks, not just on raw answer quality per dollar.

An official announcement, a model card detailing Neon’s context window and multimodal support, and published benchmarks are the next pieces to watch.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.