Get App
Download App Scanner
Scan to Download
Advertisement

Google Rolls Out Gemini 4 Neon With One Million-Token Context Window

The launch expands Google's Gemini 4 family, which also includes Carbon, designed for demanding long-running projects, and Hydrogen, built for faster everyday execution.

Google Rolls Out Gemini 4 Neon With One Million-Token Context Window
Google launches Gemini 4 Neon
Image: Google Blog

Google has unveiled Gemini 4 Neon, a new AI model built to tackle demanding tasks such as advanced coding, complex reasoning and scientific research. Aimed at enterprise users, the model can process up to one million tokens in a single context, allowing it to work with large amounts of information across text, images, audio and video.

Google DeepMind announced the launch on November 18 and stated that Gemini 4 Neon is designed to balance advanced reasoning capabilities with speed and cost efficiency. The announcement was made by Koray Kavukcuoglu, Senior Vice President at google DeepMind and Chief AI Architect at Google. 

According to the company, Gemini 4 Neon sits between Gemini 4 Carbon, intended for demanding, long-running projects, and Gemini 4 Hydrogen, which is designed for faster, everyday tasks. It is designed for complex software engineering, multi-step workflows and professional knowledge work. It can analyse lengthy documents, work with spreadsheets, interpret charts and assist with scientific and mathematical research.

Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay

ALSO READ: Nvidia Faces A New Challenge From The Very Companies Driving The AI Boom, JPMorgan Says

What is the One Million Token Context Window?

One of the key features of Gemini 4 Neon is its one million token context window. A context window determines how much information can an AI model process and refer to while completing a task. 

According to Google DeepMind's official model card, the model supports a combined input-output context window of upto one million tokens. It can process text, images, audio and video files, making it useful for analysing extensive collections of information without repeatedly splitting them into smaller sections. 

For instance, a researcher could use the model to analyse lengthy documents, examine charts and combine the findings into a structured report. 

Google states that the model supports upto 64,000 output tokens in standard synchronous responses, with longer outputs of upto one million tokens possible through background decoding.

Gemini 4 Neon: Features and Benchmark Performance

Google has highlighted the model's performance across several independent evaluations. According to its November 18 announcement, Gemini 4 Neon scored 61 on the Artificial Analysis Intelligence Index at its high thinking level, placing it third overall. It also secured 70.9% on the Vals Index.

On the coding related evaluations, the model recorded 65.4% on CursorBench 4.1 and 65.2% on Terminal-Bench 4.1. It scored 56.3% on AutomationBench, an evaluation of its ability to complete automated tasks.

For scientific and mathematical reasoning, Google reported scores of 84% on RiemannBench, 70.5% on Terminal-Bench Science and 71.3% on HLE-Diamond. On GraphWalks, which evaluates long-context reasoning across 256,000 to one million tokens, it achieved 95.2%.

Google's model card also compares Neon with competing models, including Claude Opus 5.5 and GPT-6.2 Sol. While Neon leads on several evaluations, it does not outperform competitors across every benchmark.

Gemini 4 Neon Pricing And Availability

Google has priced Gemini 4 Neon API access at $3 per million input tokens and $15 per million output tokens. Cached input tokens cost $0.15 per million, representing a 95% discount compared with standard input pricing.

The model offers four thinking levels, allowing developers to adjust reasoning intensity depending on the complexity of their tasks.

Gemini 4 Neon is rolling out through the Gemini API in Google AI Studio, Gemini for business and Google Antigravity. It is also available to Google AI Pro and Ultra subscribers through the Gemini app. Google says Neon can also work alongside Gemini 4 Hydrogen subagents in Antigravity, allowing it to coordinate multiple tasks simultaneously.

The launch expands Google's Gemini 4 family, which also includes Carbon, designed for demanding long-running projects, and Hydrogen, built for faster everyday execution.

However, Google acknowledges that Gemini 4 Neon can still generate inaccurate information. Its official model card lists March 2026 as the model's knowledge cutoff.

ALSO READ: Satya Nadella Calls Microsoft Outage Result Of 'Perfect Storm' Of Events, Warns Of AI Threats

Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices, Watch Business News Live and Latest Stories — On NDTV Profit.

Newsletters

Update Email
to get newsletters straight to your inbox
⚠️ Add your Email ID to receive Newsletters
Note: You will be signed up automatically after adding email

News for You

Set as Trusted Source
on Google Search
Listen to the latest songs, only on JioSaavn.com