Google announced Gemini 4 Neon on November 18, 2026, positioning it for complex coding, analytical reasoning, enterprise workflows, knowledge work and science. Google reported that Neon cost 32% less per task than Gemini 4 Carbon at Neon’s default high thinking level. That is a task-cost comparison, not a speed measurement. Google’s announcement also lists the model’s benchmarks, access routes and API rates.

What Google says Gemini 4 Neon is for

Google positions Neon between Gemini 4 Hydrogen, which it describes as suited to fast everyday tasks, and Gemini 4 Carbon, which it positions for the hardest, longer-running projects. Neon is aimed at complex work that needs deeper reasoning without being framed as Carbon’s most demanding, long-horizon tier.

Google says Neon has a one-million-token context window and four thinking levels. It can synthesize information from spreadsheets, PDFs, charts, audio and video within that window.

What Google’s benchmarks measure

Universe of AI’s English-language coding and rendering demonstrations

Google reported a 32% lower cost per task for Neon than Carbon at Neon’s default high setting. The comparison is tied to that setting and metric; API token rates are a separate measure.

The company also reported a score of 61 for Neon on the Artificial Analysis Intelligence Index at xhigh, ranking it third. Other results it published include 70.9% on the Vals Index, ranked second; 1,336 on the Agent Arena Index, ranked second across more than 4,200 real-world agent sessions; and 58.6 on the Coding Agent Index, ranked first overall. Google also gave Neon a 65.2% score on Terminal-Bench 4.1 and 46.2% on FrontierCode 2.0 at default high. These are results on distinct benchmarks, with different measures and settings. Google published the figures and conditions.

Compare a task-cost result with its stated baseline and reasoning setting: Google’s 32% figure uses Neon at default high and Gemini 4 Carbon.

Universe of AI’s English-language video shows creator-run coding and rendering examples alongside benchmark and pricing slides. Its demonstrations use particular prompts rather than a standardized evaluation across all tasks.

Where Google said Neon would be available—and its API rates

Google said rollout was beginning through the paid Gemini API in Google AI Studio and Android Studio, Gemini for business, Google Antigravity, and the Gemini app for Google AI Pro and Ultra subscribers. It also said Pro and Ultra subscribers could try Neon in AI Studio and Antigravity. The access routes and subscriber terms vary by service.

Google listed API rates of $3 per million input tokens, $0.15 per million cached input tokens and $15 per million output tokens. It describes the cached-input rate as a 95% discount from the standard input rate. These per-token prices differ from the reported cost per completed task. The announcement also lists a one-million-token context window.

What Google said about Gemini 4 Neon for cybersecurity

Google described a cybersecurity-focused version for finding, validating and patching vulnerabilities. In its November 18 announcement, the company said trusted defenders in the Fairwind Program would begin receiving access in the days that followed.