Google's viral AI image model 'Nano Banana' is now generally available as Gemini 2.5 Flash Image, offering new pro features for developers and enterprises.
Google has officially launched Gemini 2.5 Flash Image, the powerful AI model that went viral in August as "Nano Banana," making it generally available for developers and enterprises worldwide.
Announced on October 2, the production-ready tool is now accessible via the Gemini API, Google AI Studio, and Vertex AI.
This major release introduces new creative capabilities, including support for 10 aspect ratios and enhanced character consistency for more realistic edits.
The move solidifies Google's challenge to competitors like OpenAI and Adobe, aiming to put state-of-the-art image generation into the hands of more creators.
The model's first appeared anonymously on the crowdsourced evaluation platform LMArena as "nano-banana," quickly becoming the world's top-rated image editing model.
This strategy generated significant organic buzz before its official branding was revealed.
The general availability release on October 2 also significant production-focused upgrades.
A key enhancement is the support for 10 different aspect ratios, ranging from cinematic landscape to vertical social media formats. This allows creators to tailor content for specific platforms.
The update also refines the model's core strength: "character consistency." This feature, which preserves a subject's likeness across major edits, addresses a common failure point for many AI models.
Nicole Brichtova, a product lead at Google DeepMind, noted, "we're putting capabilities that used to require specialized tools into the hands of everyday creators, and it's been inspiring to see the explosion of creativity this has sparked."
The model's native multimodal capability allows it to process text and images simultaneously. This means it can understand an existing image and incorporate it into its creative process, rather than just generating from a text prompt.
This enables more precise and consistent edits over a conversation.
Early adopters are already integrating these capabilities. AI startup Cartwheel found the model uniquely capable of handling complex poses from any camera angle.
Co-founder Andrew Carr lauded Google, saying, "Other models couldn't render characters from arbitrary camera angles or maintain faithfulness to a pose without sacrificing 'world knowledge'. The new Gemini 2.5 Flash Image model was the first that could provide both."
Google has set the pricing at $0.039 per image and $30 per million output tokens, a competitive rate aimed at driving enterprise adoption through its Vertex AI platform.
The launch is a calculated response to a fiercely competitive market. The pressure intensified after OpenAI integrated its GPT-4o image generator directly into ChatGPT, driving a massive surge in user engagement.
Google's strategy targets a broad audience directly within its chat app, aiming for mass adoption.
More recently, competitive pressure is intensifying across the board. ByteDance has launched its Seedream 4.0 model as a direct challenger to "Nano Banana".
Meta has also pivoted its strategy for AI image generation, opting to license technology from Midjourney after internal setbacks.
The market is has been seeing specialized players emerge, like Black Forest Labs focusing on photorealism and Alibaba's model excelling at text rendering.
Google's push comes after previous stumbles in AI image generation.
The company faced backlash when an early version of Gemini produced historically inaccurate images of people, forcing a temporary suspension of the feature. This new launch is accompanied by more robust safety protocols.
To address the growing threat of deepfakes, Google is watermarking all generated content.
Images will include both a visible marker and an invisible, cryptographic SynthID watermark to clearly show they are AI-generated. This contrasts with the legal battles embroiling competitors like Midjourney.
Midjourney is currently facing a high-profile copyright lawsuit from Disney and Universal over its training data.
It highlights the complex legal and ethical landscape that all AI companies must navigate, making Google's proactive watermarking a significant strategic decision.
By embedding user-friendly editing tools into its flagship AI product, Google is positioning Gemini not just as a chatbot, but increasingly as a comprehensive creative engine, similar to OpenAI, which just launched its Sora 2 AI video model.
The move is a clear bet that accessibility and trust can win over mainstream users in the rapidly evolving field of generative AI.