In January 2026, OpenAI stealthily unveiled a complimentary translation tool named ChatGPT Translate. This innovative tool accommodates various input formats, including text, images, voice recordings, and files, and boasts support for over 50 languages. Its standout feature lies in its emphasis on semantic fidelity restoration and style refinement, enabling users to engage in seamless, continuous dialogues centered around the translated content.
Merely a day after OpenAI's move, Google responded by launching an open-source translation model, TranslateGemma. This model supports an impressive 55 languages and introduces multimodal image translation capabilities. It comes in three parameter variants: 4B, 12B, and 27B, ensuring versatility and ease of deployment across a range of devices, from mobile phones and laptops to cloud-based platforms. Remarkably, its performance outstrips that of even larger-scale models, paving the way for the widespread adoption and democratization of AI translation technology.
These two groundbreaking products herald a new era in AI translation, distinguished by multimodal functionality, lightweight design, and fierce open-source competition.