Resolution criteria

This market will resolve to YES if Google officially releases an embedding-specific model based on the Gemma 4 architecture, analogous to EmbeddingGemma and Gemma 3, at any point during the calendar year 2026.

Background

Gemma is a series of open-weights models developed by Google DeepMind. The Gemma 4 family was officially released on April 2, 2026, featuring four sizes: Effective 2B (E2B), Effective 4B (E4B), a 26B Mixture of Experts (MoE) model, and a 31B Dense model.

In September 2025, Google released "EmbeddingGemma," a dedicated 308M parameter embedding model based on the Gemma 3 architecture, specifically designed for on-device RAG (Retrieval-Augmented Generation) and semantic search tasks. As of the release of Gemma 4, this market tracks whether Google will follow the precedent set by the Gemma 3 generation and release a new, updated embedding-specific model utilizing the underlying architecture and technological improvements of the newly released Gemma 4.

This description was generated by AI.