Google introduces Gemini 4 Neon: 1 million-token context, 3 dollars per million input tokens
Google's blog post says Gemini 4 Neon is rolling out in the API, Gemini for business and the Gemini app; the company reports a lower cost than Carbon.
Google introduced its Gemini 4 Neon model in a blog post dated 18 November 2026. The model comes with a 1 million-token context window and is priced at 3 dollars per million input tokens and 15 dollars per million output tokens. Neon began rolling out the same day in the Gemini API, Gemini for business and the Gemini app for paid subscribers.
Google introduced a new model called Gemini 4 Neon in a blog post dated 18 November 2026. The company describes Neon as a model that brings frontier-level capabilities to a faster, lower-cost tier and is meant for demanding, long-running work in production. According to the post, the model began rolling out the same day.
- Price and context: 3 dollars per million input tokens and 15 dollars per million output tokens; 0.15 dollars per million tokens for cached input. The context window is 1 million tokens.
- Access: the paid tier of the Gemini API for developers (Google AI Studio and Android Studio), Gemini for business and Google Antigravity for enterprises, and the Gemini app for Google AI Pro and Ultra subscribers.
- Cost claim: according to Google, at its default high thinking level Neon runs at a 32 percent lower cost per task than Gemini 4 Carbon; the model has four thinking levels.
- Measurements: as reported by Google, Neon ranks second on the Vals Index with 70.9 percent and first on Zapier's AutomationBench with 56.3 percent.
Why it matters
Google notes that enterprise work lives in spreadsheets, PDFs, charts, audio and video, and says Neon can process all of these together in a single context window and turn them into structured output. The company also says it applies Carbon-level protections against misuse in cybersecurity and biosecurity while declining less legitimate work. All of these results rest on measurements that Google reports in its own post.
For companies in Türkiye that use models through an API, the announcement brings two concrete items: the per-token price and a cost per task that can be tuned with the thinking level. Teams running long documents or multi-step agent work can try the same job at different thinking levels and measure the balance of quality and cost on their own data; rankings published by a vendor do not replace results on your own work samples. For agent set-ups built for your company, see our <a href="/yapay-zeka/sirkete-ozel-yapay-zeka-ajanlari">company-specific AI agents</a> page.
Sources
More articles
Google: Russian espionage group Star Blizzard takes over cloud accounts and downloads entire company drives
The attackers got in with stolen admin credentials and repeated MFA prompts, then downloaded the whole of the organisation's cloud drive.
November 18, 2026 · Source: Infosecurity Magazine59 European tech founders and scientists ask the EU for a dedicated AI line in its 2028-2034 budget
The letter, published on 18 November 2026, warns that anything not secured in the EU's 2028-2034 budget talks will not exist before 2035.
November 18, 2026 · Source: EuronewsAnthropic prices its IPO at 83 dollars a share; stock to trade on Nasdaq under the symbol ANT
The company says the offering consists of 980 million Class A shares. Renaissance Capital notes the price came in above the 70 to 78 dollar range.
November 18, 2026 · Source: Anthropic