SIAgentNews.com
← All news
DevelopingGemini

Google prepares new Gemini 4 Neon model for paid users

Code references and pricing for a new Gemini 4 Neon model suggest a high-performance option for developers and paid subscribers.

· First reported by TestingCatalog AI News

$3price per million input tokens

The short version

  • Gemini 4 Neon appeared in the Gemini web app upsell section and the Google Cloud SKU catalog.
  • The model is priced at $3 per million input tokens and $15 per million output tokens for text.
  • A new variant for the Antigravity CLI suggests Neon may function as a 'Pro' orchestrator model.
  • Google employees have reportedly described the upcoming model as a 'daily driver'.

Why it matters

For developers, Neon provides a new option for building complex agents on Google Cloud. For paid subscribers, it signals a move toward more powerful models for everyday heavy lifting.

Evidence of a new release

Google is currently preparing a new addition to its AI lineup, with Gemini 4 Neon emerging as the most likely candidate. While the company has not made an official announcement, several technical breadcrumbs have appeared over the last few days. These signs suggest a coordinated rollout for a model intended for broader use.

The model first surfaced as a code reference in the Gemini web app on November 16. Specifically, the identifier 'gemini-4-neon' appeared in the upsell section of the site. This placement indicates that the model will likely be reserved for users who pay for a subscription rather than being available for free.

Further confirmation arrived on November 18 when the model appeared in the Google Cloud SKU catalog. This catalog lists the specific technical specifications and costs for different models. The appearance in the cloud catalog follows a pattern Google has used for previous releases, where the model is listed before it is officially opened to the public.

Pricing and capabilities

The Google Cloud entries provide specific details regarding how the model will be billed for developers. For text processing, the model is listed at $3 per million input tokens and $15 per million output tokens. There is also a discounted rate of $0.15 per million tokens for cached input, which helps reduce costs for repeated data processing.

The model is designed to handle a variety of different media types beyond just standard text. The catalog lists separate SKUs for image, video, and audio inputs. This versatility suggests that Neon is built to be a multi-modal tool, capable of understanding and generating various forms of digital content in a single ecosystem.

While early speculation suggested Neon might be a smaller, faster model like a 'Flash Lite' version, the pricing and labels suggest otherwise. The inclusion of a 'Pro' label in some contexts points toward a larger, more capable model. This would position Neon as a high-performance option rather than a budget-friendly, low-latency alternative.

The role of an orchestrator

New information from the Antigravity CLI software provides a glimpse into how Neon might function in complex workflows. A new variant called 'gemini-pro-orchestrator-neon' was added to the software's latest build. An orchestrator is a specific type of AI setup that plans out multi-step tasks and then hands those individual steps off to other models to complete.

In this specific setup, the software pairs different models to handle different parts of a job. For example, it uses Gemini 4 Carbon for the planning phase and Gemini 4 Hydrogen for the actual execution. By using Neon as an orchestrator, Google is signaling that this model is particularly good at high-level reasoning and task management.

This architecture allows for more sophisticated 'agentic' behavior, where the AI can break down a large goal into smaller pieces. Instead of trying to do everything at once, the orchestrator ensures that the right model is used for the right part of the job. This makes it a powerful tool for developers who want to build autonomous systems that can complete complex projects.

Positioning in the lineup

Google is rapidly expanding the Gemini 4 family, with new models appearing in the API and various products week after week. Gemini 4 Hydrogen is already integrated into Google Search and Tasks, while Gemini 4 Carbon and Argon serve as the larger, more powerful models in the current set. Neon is being positioned to fill a specific gap in this growing ecosystem.

Internal sources at Google have reportedly referred to the upcoming model as a 'real daily driver.' This phrasing suggests that Google wants Neon to be the primary model for standard, heavy-duty work. It is intended to be the reliable workhorse that users turn to for their most frequent and demanding interactions.

The release of Neon comes alongside rumors of other models like Gemini Helix and Gemini Krypton. There is also a long-awaited upgrade for the Nano Banana model based on Gemini 4. However, the 'daily driver' framing suggests that Neon is the priority for the immediate future. It aims to provide a robust option for both paid subscribers and developers building agents on Google Cloud.

What to watch

  • When Google officially announces the release date for Gemini 4 Neon.
  • Whether the model will be available on the Vertex AI platform as well as the standard API.
  • How the 'orchestrator' capabilities of Neon perform compared to existing models like Gemini 4 Carbon.

Try it

Check the Google Cloud SKU catalog for 'gemini-4-neon' to see the current listed pricing for text and image inputs.

Sources and links

First reported by TestingCatalog AI News. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news