International Edition
Latest News
Technology

Google Launches Gemini 4 Neon AI Model at Lower Cost, Available Now

Google launched Gemini 4 Neon on November 18, positioning the model between its flagship Carbon and speed-focused Hydrogen tiers with a pitch that it matches frontier reasoning at a fraction of the cost. The model is available immediately…

Google Launches Gemini 4 Neon AI Model at Lower Cost, Available Now

Google launched Gemini 4 Neon on November 18, positioning the model between its flagship Carbon and speed-focused Hydrogen tiers with a pitch that it matches frontier reasoning at a fraction of the cost.

The model is available immediately to Google AI Pro and Ultra subscribers in the Gemini app, Google AI Studio, Android Studio, Vertex AI and the company’s new Antigravity agent platform, the company said. Pricing is set at $3 per million input tokens and $15 per million output tokens, with cached input discounted to $0.15 per million tokens.

Google’s own benchmarks place Neon third on the Artificial Analysis Intelligence Index with a score of 61, ahead of Anthropic’s Claude Opus 5.5 at 59 and OpenAI’s GPT-6.2 Sol at 57. Neon also leads on the Vals Index at 70.9 percent and AutomationBench at 56.3 percent, though it trails Opus 5.5 on Terminal-Bench 4.1 and FrontierCode 2.0 and trails GPT-6.2 Sol on Chartography, according to figures the company released. Thai Tran, AI product lead at Glean, said in Google’s announcement that early evaluations showed Neon “matched or exceeded Gemini 4 Carbon on overall quality at 3x the speed and 60% of the cost.” Google did not publish absolute completion times in seconds for matched workloads. The model supports a 1-million-token context window and four configurable thinking levels, with “high” set as the default. At that setting Google says Neon delivers Carbon-level intelligence scores at 32 percent lower cost per task and 50 to 80 percent less per task than competing frontier models. An “xhigh” thinking mode pushes the Intelligence Index score to 61 and leads benchmarks including SWE-sweep at 39.0 percent and GraphWalks at 95.2 percent while costing over 36 percent less per task than Opus 5.5 and GPT-6.2 Sol at their maximum settings, according to Google. Neon replaces Argon in the Gemini app model picker, where it now appears under a “Deeper analysis” heading while Hydrogen sits under “Everyday help.” The app can process spreadsheets, PDFs, charts, audio and video in a single prompt and draws context from Gmail and Docs through Gemini Tasks.

Inside Antigravity, Neon acts as a high-judgment Orchestrator that plans architecture and reviews code while delegating rapid edits and test runs to parallel Hydrogen subagents, Google said. A demonstration showed a Neon agent swarm reconstructing an image across 100,000 spreadsheet cells using more than 500,000 live formulas to compute 3D depth maps without Apps Script. On security, Neon scored 70.7 percent on CWE-Bench v2 for vulnerability patching and recorded a 0.0 percent attack success rate on Gray Swan’s indirect prompt injection evaluation, with access rolling out to vetted defenders via the Fairwind Program, Google said. VentureBeat said that Neon’s headline API rates are 50 percent higher than Anthropic’s mid-range Claude Sonnet 5.5 and OpenAI’s mid-range GPT-6.2 Sol, but Google reports Neon beats both on the Intelligence Index at its highest reasoning setting. Neon also undercuts the larger Opus 5.5’s standard rates of $4 input and $20 output by 25 percent while scoring within about a percentage point on two engineering benchmarks.

Early user examples cited by Google and VentureBeat include tracing unexpected charges across billing records, auditing unfamiliar codebases for defects, testing machine-learning hypotheses and building inspectable spreadsheets. Thoughtful Lab founder Karina Nguyen said Neon helped her team understand complex software architectures quickly and spotted subtle vulnerabilities other models missed. Cartwheel co-founder and chief scientist Andrew Carr described using Neon to examine prior research, test hypotheses and incorporate feedback as experiments progressed.

“Neon is built for complex coding, analytical reasoning, and production-grade enterprise workflows,” Koray Kavukcuoglu, SVP of Google DeepMind and Chief AI Architect at Google, said in a statement to The Hindu. “While Carbon tackles your hardest, long-horizon projects, and Hydrogen handles fast everyday execution, Neon balances depth and efficiency.” Google said Neon maintains Gemini 4 Carbon-level safeguards against severe misuse and prompt injection and includes advanced model alignment. The model is listed in Vertex AI documentation for verified enterprise customers in global, US and EU regions.

About the author: Technology Desk

Technology Desk at Archynewsy.