Google is rolling out Gemini 4 Neon in the Gemini app for AI Pro and Ultra subscribers, starting November 18. Neon takes the model picker’s Deeper analysis slot in place of Argon. Gemini 3.1 Pro is gone from the app.
The key detail for paying users is that Neon is not reserved for Ultra. AI Pro, at $19.99 a month in the US, gets it too.

What changes in the Gemini app model picker
According to 9to5Google, the picker now offers four choices:
- Auto: adapts to your needs; Gemini 3.5 Flash-Lite now sits behind it for paid users
- Everyday help: Gemini 4 Hydrogen
- Deeper analysis: Gemini 4 Neon
- Code & complex work: Gemini 4 Carbon
Neon comes with a 1M-token context window. It also offers four thinking levels: Low, Medium, High and Extra High.
Dropping 3.1 Pro means anyone who relied on it now has to move to a Gemini 4 model. As of November 19, we haven’t seen Google spell out Neon’s usage caps for each plan. The same goes for any free-tier access.
What Google claims for Gemini 4 Neon
Koray Kavukcuoglu, SVP of Google DeepMind, introduced the model on Google’s blog:
Today, we’re introducing Gemini 4 Neon — bringing frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows.
Google backs that up with benchmark rankings of its own choosing. It says Neon places third overall on the Artificial Analysis Intelligence Index, scoring 61 at its highest thinking level. Google also cites second place on the Vals Index at 70.9%. It claims second place for steerability on Arena’s Agent Arena Index as well. These are Google’s numbers, not results androidpure has tested.
Gemini Tasks can now run on Neon
9to5Google reports that Tasks in Gemini can now use Neon. That lands shortly after Google made Gemini Tasks free in the US.
Gemini 4 Neon API pricing and availability
Developers can use Neon through the Gemini API and AI Studio. It is also rolling out to Gemini for business. Paid Pro and Ultra users get it in Google Antigravity too.
- Input: $3 per 1M tokens, or $0.15 cached
- Output: $15 per 1M tokens
- Context window: 1M tokens
9to5Google notes that this is twice what Hydrogen costs. Argon and Carbon cost $2 for input and $10 for output per 1M tokens until November 30. After that, they rise to $4 and $20. From December, Neon will cost less per token than either of them.
Sources: Google, 9to5Google, Android Authority






