Powered by Gemini models, the update enables Translate to better handle context, idioms, and informal language. Rather than presenting a single result, the app now offers alternative phrasings and explanations of tone and nuance. Users can see why a translation was generated and refine the output through follow-up prompts.
The change reflects a broader evolution in AI translation UX — from static single output toward guided iterative refinement. Earlier this year, OpenAI launched ChatGPT Translate, a standalone interface that lets users adjust tone, audience, and style through presets and iterative prompts rather than relying on a single output.
Together, these developments suggest that leading AI providers are moving toward a more interactive translation model, where users shape results instead of simply receiving them.
Accuracy remains critical, but context, clarity, and control are increasingly becoming part of user expectations. For the language industry, this may signal growing demand for systems that combine strong baseline quality with built-in refinement and guidance features.
Nano Banana Update
The update also fits within Google’s broader AI strategy. On the same day, the company introduced Nano Banana 2, an image generation model capable of translating and localizing text within generated images — a capability now broadly available across the Gemini app, Google Search/Lens, and other AI tools. ​​The move extends translation functionality beyond text-based interfaces and into multimodal AI workflows.
