Google is broadening the range of Gemini large language models accessible to developers on its Vertex AI platform. The company has made Gemini 1.0 Pro generally available after a period in public preview. Additionally, Gemini 1.0 Ultra is now generally available “via allowlist.” Furthermore, Google has announced Gemini 1.5 Pro, an update to the existing Gemini Pro model, which is capable of handling a context of one million tokens and is currently in private preview.
Key Takeaway
Google is expanding the availability of Gemini large language models for developers on its Vertex AI platform, making models like Gemini 1.0 Pro and Gemini 1.0 Ultra generally available, and introducing new tools and integrations to enhance the developer experience.
Support for Adapter-Based Tuning and Developer Tools
Google is introducing support for adapter-based tuning in Vertex, with reinforcement learning from human feedback and distillation techniques coming soon. This will enable developers to more easily augment their models with up-to-date data and connect the Gemini model to external APIs. Moreover, developers can now access the Gemini API from the Dart SDK, facilitating its use in Dart and Flutter apps. Google is also streamlining the use of the Gemini API with Project IDX and integrating it with Firebase through an extension, enhancing the capabilities for developers.
