Skip to content
Theme
OnDevice Tools

AIAI Models & Research

Google releases Gemini 4 Neon for coding and science work

Google has launched Gemini 4 Neon, a multimodal model aimed at coding, knowledge tasks and science. The Decoder reports that it costs more than earlier Gemini 4 models and offers a 1 million token context window.

Illustration of Google Gemini 4 Neon announcement showing model pricing and context window

Key takeaways

  • Google released Gemini 4 Neon through an API on 18 November 2026, according to The Decoder.
  • Neon handles text, images, video, audio and PDFs, with a context window of about 1 million tokens.
  • It costs $3 for 1 million input tokens and $15 for 1 million output tokens.
  • Benchmarks are still pending, and the model cannot generate images or audio.
In this article
  1. Launch details
  2. Capabilities and limits
  3. Pricing and model lineup
  4. Four models in four weeks
  5. What it means for developers

Launch details

Google's Gemini 4 Neon model is now available as a multimodal AI system for coding, knowledge tasks and scientific projects. The Decoder reported on 18 November 2026 that the model is live through an application programming interface, or API, which lets other software use it. It is the fourth new model in the Gemini 4 line within four weeks.

Google says Neon is a frontier model designed for complex work and fast task completion. The model can process text, images, video, audio and PDF files. Its context window is about 1 million tokens, and it can produce up to 1 million output tokens.

The report focuses on the API release. It does not mention a consumer product launch. No wider release date is given.

Capabilities and limits

A context window is the amount of material a model can consider at one time. Neon's large limit means it can accept a large amount of input before producing output. Tokens are the units counted for input and output.

Multimodal means the model can work with more than text. Neon's support for PDFs and other media may help with tasks that involve documents and mixed inputs.

Neon supports Computer Use, code execution, file search and grounding with Google Search and Maps. Grounding links responses to external information. These features let the model connect to external tools and services. It cannot generate images or audio.

The Decoder says performance benchmarks are expected soon. Until those results appear, Neon's real-world performance remains unconfirmed.

Pricing and model lineup

Neon is available at $3 for 1 million input tokens and $15 for 1 million output tokens. That is higher than the introductory prices for Carbon and Argon, which are $2 for input and $10 for output. It is also double Hydrogen's $1.50 input and $7.50 output prices.

The Decoder says Carbon and Argon are due to rise to $4 for input and $20 for output on 1 December 2026. At the current introductory rate, Neon costs $1 more for input and $5 more for output. After the rise, Carbon and Argon would cost $1 more than Neon for input and $5 more for output.

The report does not say whether Neon's price will change. The separate input and output charges may matter for teams that send large volumes of data to a model. Workloads with long outputs can cost more because output tokens are priced higher.

Four models in four weeks

Illustration: AI, Gemini logo, Google logo, environment and the labels Gemini 4, Gemini 4 Neon, Google

Neon is the fourth Gemini 4 model Google has shipped in four weeks. The Decoder names Carbon, Argon and Hydrogen as earlier models in the range. The quick sequence shows Google is adding several models in a short period.

The material does not list every Gemini 4 model or provide a full release schedule. It also leaves unclear whether Neon will replace or sit alongside Carbon, Argon and Hydrogen.

The report does not confirm how Neon compares with rival models from OpenAI or Anthropic. The Decoder presents that comparison as a possibility, not a confirmed fact. Any such comparison should be treated as speculation until benchmarks appear.

What it means for developers

For developers, Neon adds another high-end option in Google's Gemini 4 family. Its support for multiple media types and a large context window may suit projects that involve code, documents and external data. Features such as code execution and file search add tool-like capabilities.

Cost-sensitive teams may compare Neon with Hydrogen, which is cheaper and may be enough for some tasks. They should also weigh Neon's lack of image and audio generation against their needs.

Google's documentation contains full specifications for Neon. The Decoder says test results are still pending. Teams evaluating the model should check cost, context limits and supported features against their own needs.

Sources

  1. Google ships Gemini 4 Neon, its fourth new Gemini model in a monthThe Decoder

Tags

Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.