Key points
- Google DeepMind released Gemini 4 Neon, its fourth Gemini 4 model in roughly four to seven weeks, depending on the outlet counting.
- Neon is priced at $3 for input and $15 for output per million tokens, above Carbon and Hydrogen.
- Neon is reaching paid API users, Gemini for business and Antigravity, with Cursor and GitHub Copilot support on day one.
- Early tests put Neon second on Agent Arena and the Vals Index, though capacity errors showed up in one set of Antigravity sessions.
Why it matters for marketers: Teams building agent workflows now have a higher-tier Google model to test against rival options, with cost per task and capacity limits as the practical tradeoffs.
Google DeepMind has released Gemini 4 Neon, a higher-tier model aimed at software engineering, knowledge work and science tasks. It is the fourth Gemini 4 model in a short span: The Decoder says four weeks, while Latent Space says about seven. Latent Space also reports that Argon, released roughly 1.5 months ago, is now retired.
Neon costs more per token than Carbon's and Hydrogen's rates, but it has a context window of about one million tokens and handles text, images, video, audio and PDFs. It does not generate images or audio. Google is rolling it out through the paid Gemini API, Gemini for business and Antigravity, and Pro and Ultra users can pick it in the Gemini app as "Deeper analysis."
Early third-party results are strong but mixed. Latent Space cites Artificial Analysis, Arena and Vals placing Neon at or near the top of several agent and coding rankings, with Vals putting it second at lower cost per task than the leader. Cursor reported a gain on its own benchmark over Hydrogen. In 13 early Antigravity sessions, one tester logged 57 capacity errors and four stalled runs.
For marketing and content teams, the practical question is fit and cost. One developer quoted by Latent Space is using Neon for copy and other models for implementation, a sign that teams are splitting work across models. Compare cost per task, not just token price, and check whether capacity holds up before moving production workflows. Watch for Google's full benchmark release and whether Neon becomes the default replacement for older Gemini tiers.
Sources
This brief is our own summary. Read the full reporting at the source.
