Google DeepMind Unveils 'Gemini 2.5 Flash' for Radical Image Editing
Google DeepMind introduces 'Gemini 2.5 Flash Image Generation', a new image editing model now available in the Gemini app. This model significantly enhances image manipulation, offering radical alterations while maintaining recognizable subjects.
The Gemini 2.5 Flash model, available via the Gemini API, Google AI Studio, and Vertex AI, costs $30 per million output tokens. It builds on the Gemini language model's image generation, behaving accurately in prompt implementation, akin to GPT-4.
One key feature is character consistency, ensuring people, objects, or animals remain consistent across different images. The model also excels in stylistic transformations, transferring color, texture, or design from one object to another while retaining its form and details.
Users can now make precise, locally limited edits via text input, such as blurring backgrounds, removing spots, adding colors, or deleting objects, without manual selection tools. Moreover, it merges up to three images for complex compositions with a single prompt, demonstrating real-world reasoning and understanding of simple causal relationships.
To access these features, users need to switch to the 'Flash' language model within the Gemini app.
Google's new Gemini 2.5 Flash Image Generation model brings advanced image editing capabilities to the Gemini app. With radical image alterations, character consistency, precise edits, and complex compositions, it offers users a powerful tool for creative expression. The model is available now via the Gemini API, Google AI Studio, and Vertex AI, with usage costs of $30 per million output tokens.
