Artificial intelligence news.
Straight from the original announcements.

Google DeepMind launches Gemini 4 Neon for complex agent workflows

November 18, 2026, 22:35By the AI News Bites Editorial TeamSource: Google DeepMind
Top storyGoogle DeepMind launches Gemini 4 Neon for complex agent workflows
Image: Google DeepMind

Google DeepMind has introduced Gemini 4 Neon, a model designed for fast, end-to-end completion of complex software engineering, knowledge work and scientific workflows. The company says Neon combines frontier-level intelligence with lower costs, steerable agent behavior and a 1 million-token context window.

Key points
  • Gemini 4 Neon is designed for software engineering, enterprise work and science workflows.
  • It has a 1 million-token context window and four thinking levels.
  • Pricing starts at $3 per million input tokens and $15 per million output tokens.
  • The model is rolling out across Google AI Studio, Gemini for business, Google Antigravity and the Gemini app.
  • A cyber version is planned for trusted defenders through the Fairwind Program.

Gemini 4 Neon targets long, complex tasks

Google DeepMind says Gemini 4 Neon ranks third overall on the Artificial Analysis Intelligence Index, with a score of 61 at its xhigh thinking level. It ranks second on the Vals Index and second for steerability on the Agent Arena Index, which measures how well a model adapts to user corrections during a task.

The model is built to combine spreadsheets, PDFs, charts, audio and video in a single context window, then turn those inputs into structured deliverables. Google says Neon ranked first on the Legal Research Bench and Tax Agent Bench, and achieved first place on Zapier’s AutomationBench with a score of 56.3%.

For scientific and mathematical work, the company reports first-place results on Terminal-Bench Science at 70.5%, RiemannBench at 84.0% and HLE-Diamond at 71.3%. These are company-reported benchmark results.

Built for coding and production workflows

Neon is designed for long-running, tool-heavy agent sessions. Google DeepMind says it can reduce redundant file reads, share progress during work and handle context compaction in extended sessions. It can also act as an Orchestrator directing parallel Hydrogen subagents in Google Antigravity.

The model ranks first overall on the Artificial Analysis Coding Agent Index, with a score of 58.6. It scored 65.2% on Terminal-Bench 4.1 and 46.2% on FrontierCode 2.0 at its default high setting, according to Google.

Google also highlights multi-file debugging and repository-wide refactoring. On SWE-sweep, Neon scored 12.8% on the mini-SWE harness and 39.0% in Google Antigravity. The company says its agent swarms have also created interactive 3D experiences, a large formula-driven Google Sheet and other software projects.

Pricing, availability and safety controls

Gemini 4 Neon costs $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens. Google says the cached input price represents a 95% discount. Neon has four thinking levels, allowing teams to adjust intelligence and token spending for different workloads.

At its default high thinking level, Google says Neon has a 32% lower cost per task than Carbon. The company also reports roughly 50% to 80% lower cost per task than top competing frontier models on the Vals Index and Agent Arena at high, and says the xhigh setting costs 36%+ less per task than competing models at their highest setting.

The model is rolling out in the Gemini API in Google AI Studio, Gemini for business through the API and model dropdown, Google Antigravity for paid Google AI Pro and Ultra subscribers, and the Gemini app for Pro and Ultra subscribers.

Google says Neon maintains Gemini 4 Carbon-level protections against severe misuse and prompt injection. Its cyber version is due to roll out to trusted defenders in the coming days through the Fairwind Program. Google reports that Neon for cyber achieved 76% on the hard tier of AutoBaxBench and 90.5% on SEC-bench Pro.

Source
Google DeepMind: Introducing Gemini 4 Neon

Related news

Google DeepMind introduces Gemini 4 Hydrogen for faster agentic workTop story
ModelsNovember 12, 2026, 19:00

Google DeepMind introduces Gemini 4 Hydrogen for faster agentic work

Google DeepMind has introduced Gemini 4 Hydrogen, a model designed for rapid execution across everyday coding, agentic workflows and consumer tasks. The company says…

Google makes Gemini 4 Carbon available to developers and consumersTop story
ModelsNovember 10, 2026, 17:03

Google makes Gemini 4 Carbon available to developers and consumers

Google DeepMind has made Gemini 4 Carbon available to developers, enterprises, scientists and consumers, positioning it as the company’s strongest model for complex…

Google tests cheap classifier to route Gemini tasks and cut agent costs
For developersNovember 3, 2026

Google tests cheap classifier to route Gemini tasks and cut agent costs

Google's developer blog argues that cost per successful task, not price per million tokens, now decides the economics of coding agents. In a 50-question benchmark, a…