Gemma 4
A compact open-weight model for fast reasoning, classification, and everyday analysis.
Gemma 4 · Open weights
Overview
Gemma 4 is Google's open-weight model family, designed to bring useful reasoning and multimodal capabilities to a wide range of hardware and deployment sizes.
In Aether, the 31B instruction-tuned variant is a practical fast option for lightweight questions, extraction, and first-pass analysis.
+Pros
- Very low-latency responses for smaller tasks
- Open-weight family with efficient deployment options
- Reasoning and multimodal support across the Gemma 4 family
-Cons
- Less suitable for complex coding or long, high-stakes analysis
- The Aether entry is a focused instruction variant rather than the whole Gemma family
Best suited to
Short questions and quick explanations
Classification and information extraction
Lightweight document analysis
Fast first drafts and feedback summaries
Aether Workspace Position
Available as a lightweight model in Aether.
Aether can use Gemma 4 for short factual prompts, classification, extraction, and other tasks where speed matters more than maximum depth.
Research sources
- Gemma 4 model overviewGoogle AI for Developers
- Gemma 4 model cardGoogle AI for Developers
