Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

blog.google.9to5google.com.cryptobriefing.com+1.Google on November 18 released Gemini 4 Neon, a new model for complex software engineering, knowledge work and science workflows. It costs $3 per million input tokens and $15 per million output tokens and has a 1 million-token context window, according to Google's announcement. Cached input tokens cost $0.15 per million, a 95% discount. Neon is meant to sit between Google's flagship and its everyday model. Google says it reaches near-flagship results at lower cost and higher speed.blog.google
Neon is the latest addition to a Gemini 4 lineup that already includes Argon, Carbon and Hydrogen. 9to5Google reported that Neon comes one week after Carbon and Hydrogen, and that it replaces Argon. In the Gemini app, Neon appears as "Deeper analysis" in the model picker, alongside Hydrogen for "Everyday help" and Carbon for "Code & complex work." Gemini 3.1 Pro is being dropped from the picker.9to5google.com
Google DeepMind executive Koray Kavukcuoglu wrote that "Carbon tackles your hardest, long-horizon projects and Hydrogen handles fast everyday execution," while Neon "balances depth and efficiency," according to the Google blog post. Neon costs twice as much as Hydrogen, 9to5Google noted. It will be 25% cheaper than Carbon once Carbon's introductory price of $2 per million input tokens ends on November 30. After that date, Carbon's price rises to $4 per million input tokens and $20 per million output tokens.blog.google+1
The model is rolling out in several places at once. These include the Gemini API in Google AI Studio, Gemini for business, and Google Antigravity for paid AI Pro and Ultra subscribers. It is also coming to the Gemini app for Pro and Ultra subscribers, Google said.blog.google
Google says Neon ranks third overall on the Artificial Analysis Intelligence Index, scoring 61 at its highest "xhigh" thinking level. It ranks second on the Vals Index with 70.9%. Google also reports first place on the Artificial Analysis Coding Agent Index, with a score of 58.6, and on Zapier's AutomationBench. At its default "high" setting, the company says Neon costs 32% less per task than Carbon. These figures come from Google's own launch materials.blog.google
On Arena's new Agent Arena Index, Neon debuted in second place with a score of 1336, based on more than 4,200 agent sessions. Only Anthropic's Claude Fable 5.2 ranked higher, Crypto Briefing reported.cryptobriefing.com
Google is also preparing a cybersecurity version of the model, called Gemini 4 Neon for cyber. It will reach "trusted defenders in the coming days as part of the Fairwind Program," according to the company. Eligible members can request access through their Google Cloud account teams. Google says this version scored 90.5% on SEC-bench Pro for finding vulnerabilities. It also says the model scored 76% on the hard tier of AutoBaxBench, which tests secure code generation, and that this matches the best result recorded on that tier.blog.google