GEMPIX2 (Nano Banana 2): Google's Next-Generation Image Model Coming Soon?

By: Banana Bro | Published: November 5, 2025 | Reading time: approx. 6 minutes

Following Google's official launch of Gemini 2.5 Flash Image (codenamed: Nano Banana) in August 2025, its powerful image generation and editing capabilities created quite a buzz. Now, the industry is buzzing with frequent reports about the next-generation model—GEMPIX2 (or Nano Banana 2). What exactly is it? How is development progressing? When will it be released? This article brings you the latest updates.

What is GEMPIX2?

"GEMPIX2" is currently believed to be Google's new-generation AI image generation model in development. According to multiple sources, it's very likely the successor to Gemini 2.5 Flash Image, belonging to an extension of the "Nano Banana" series.

Industry analysts speculate that the GEMPIX2 name derives from "Gemini + Pixel," suggesting it may not just be an AI model update, but an important milestone in deeply integrating AI generation technology into Pixel devices.

Latest Developments: Internal Previews Emerge

Preview Card Discovery

According to TestingCatalog's findings, a preview card named "Nano Banana 2" recently appeared in the Gemini app UI. This is considered a precursor to GEMPIX2.

Developer Evidence

Developers have discovered suspected new model identifier references in Gemini's web configuration files. While not yet available, this typically indicates Google is in the internal testing phase.

Potential New Features

While Google hasn't released any official documentation, based on leaks and analysis, GEMPIX2 may bring improvements in the following areas:

Higher Fidelity Image Generation

Compared to Gemini 2.5 Flash (Nano Banana), GEMPIX2 may support richer texture details and more precise lighting modeling.

Enhanced Multi-Character Consistency

Expected to further improve character consistency performance across multiple images or scenes.

On-Device Inference Optimization

"Pixel optimization" is considered one of the core goals of the GEMPIX series, meaning local generation tasks on Pixel phones will be more fluid.

Semantic-Level Editing Capabilities

Support for "semantic modifications" of images under natural language instructions, such as:

• "make this person smile"
• "change to nighttime lighting"
• "replace background with beach"

When Will It Be Released?

Currently, there's no official confirmation of GEMPIX2's release date. However, multiple information sources (such as TestingCatalog) indicate that the model may officially launch on the Gemini platform within the coming weeks, or debut at the next Made by Google event.

Estimated Timeline: Mid-November to Early December 2025

This means we may soon see GEMPIX2 in actual applications, especially on Pixel 9 or Pixel Fold series devices.

Project Summary

Item	Details
Model Codename	GEMPIX2 / Nano Banana 2
Series	Gemini 2.5 Flash Image successor
Status	Internal testing phase (preview appeared)
Expected Release	November-December 2025
Core Highlights	High-precision image generation, semantic editing, on-device optimization

Author's Perspective

From "Nano Banana" to "GEMPIX2," Google is gradually connecting the generation pipeline for text, images, and video. If GEMPIX2 truly possesses on-device optimization capabilities as rumored, it may signal that the era of real-time and localized AI image generation is about to arrive.

---

Stay tuned for more updates on GEMPIX2 and Nano Banana 2!