According to @GoogleDeepMind, a new addition to the Gemma open models family called Gemma 3 270M was announced as a tiny yet capable AI designed for task-specific fine-tuning with built-in instruction following; source: https://twitter.com/GoogleDeepMind/status/1956393664248271082. The announcement provides a direct build link for developers to get started with the model, highlighting immediate availability for integration and experimentation; source: https://twitter.com/GoogleDeepMind/status/1956393664248271082. The X post itself does not include performance benchmarks, pricing, or licensing details within the tweet text, focusing instead on positioning the model for fine-tuning and instruction-following use cases; source: https://twitter.com/GoogleDeepMind/status/1956393664248271082.

According to Sam Altman, the open model will be accessible to all services, regardless of the number of monthly active users, ensuring broad usage potential for various platforms.