Ateron/Gemma-4-MoonGem-31B

VISIONPricing:Input $0.48 / Cached $0.1 / Output $1.44Concurrent Unit Cost:2Model Size:31BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 3, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Ateron/Gemma-4-MoonGem-31B is a 31 billion parameter language model from the Gemma-4 family, created by Ateron. This model is a result of a multi-phase merge using Mergekit, combining several base models to achieve its capabilities. With a context length of 32768 tokens, it is designed for general language generation tasks.

Loading preview...

Overview

Ateron/Gemma-4-MoonGem-31B is a 31 billion parameter language model built upon the Gemma-4 architecture. This model was developed by Ateron through a sophisticated multi-phase merging process using Mergekit, integrating various base models to enhance its performance and capabilities. It supports a context length of 32768 tokens, making it suitable for processing longer inputs and generating more extensive outputs.

Mergekit Configuration

The model's creation involved a three-phase merging strategy:

  • Phase 1 (Bleed): Utilized the ties merge method, combining Gemma-4-Glimmer, Gemma-4-Gutenberg, and Gemma-4-Gemopus with Gemma-4-Scotoma-V2 as the base model. This phase involved specific density and layer-wise weighting parameters.
  • Phase 2 (Crimson): Employed the task_arithmetic merge method, integrating Gemma-4-Melinoe-VL, Gemma-4-MeroMero-V2, and Gemma-4-MusicaV1 with Gemma-4-Scotoma-V2 as the base.
  • Phase 3 (Blood): Also used the task_arithmetic merge method, combining Gemma-4-MicroMix-P1 and Gemma-4-MicroMix-P2 with Gemma-4-Scotoma-V2 as the base.

Key Characteristics

  • Architecture: Based on the Gemma-4 family.
  • Parameter Count: 31 billion parameters.
  • Context Length: 32768 tokens.
  • Development Method: Advanced multi-phase merging using Mergekit, indicating a focus on combining strengths from multiple specialized models.

Intended Use Cases

Given its large parameter count and substantial context window, MoonGem-31B is suitable for a broad range of natural language processing tasks, including but not limited to:

  • Complex text generation
  • Detailed content creation
  • Advanced conversational AI
  • Summarization of long documents
  • Question answering over extensive texts