Nimbz/Gemma-4-Gembrain-X-Core-31B

VISIONConcurrency Cost:2Model Size:31BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 30, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

Nimbz/Gemma-4-Gembrain-X-Core-31B is a 31 billion parameter merge model based on the Gemma 4 architecture, designed for enhanced creative writing, roleplay, and strong reasoning capabilities. It features a unique dual-mode operation for fast general inference or deep, prompt-adherent reasoning, and excels at generating varied and descriptive image prompts. The model maintains uncensored responses and high prompt adherence, with a context length tested up to 30,000 tokens.

Loading preview...

Model Overview

Nimbz/Gemma-4-Gembrain-X-Core-31B is a 31 billion parameter merge model built upon the Gemma 4 architecture, representing a ground-up rebuild of the original Gembrain-31B. Developed by Nimbz, this model aims to improve upon its predecessor by enhancing creative writing, roleplay, and reasoning while maintaining prompt adherence and an uncensored nature. It integrates strengths from multiple base models, including those focused on reasoning, prose variety, character agency, and the original Gemsicle identity.

Key Capabilities

  • Dual-Mode Operation: Offers a "non-thinking" mode for fast, versatile general use, light roleplay, and image prompt generation, and a "thinking on" mode for deeper understanding, strict character adherence, and complex reasoning.
  • Creative Roleplay & Swipe Variety: Excels in open-ended roleplay scenarios, providing genuinely different angles and varied responses without repetition. Characters pursue their own agendas and maintain consistency with detailed descriptions.
  • Uncensored & Prompt-Adherent: Designed to embrace explicit content with vivid descriptions rather than euphemisms, while strictly adhering to user prompts and character traits.
  • Image Prompt Generation: Capable of producing clean, vivid, and distinct image prompts, often generating multiple coherent options in a single response.
  • Context Awareness: Preserves global-attention layers, ensuring reasoning and detail-tracking remain intact across long chats, tested up to 30,000 tokens.

Ideal Use Cases

  • Creative Writing & Roleplay: For users seeking dynamic, varied, and character-driven narrative generation.
  • Complex Reasoning Tasks: When detailed, coherent, and strictly adherent output is required, especially with the "thinking on" mode.
  • Image Prompt Generation: For artists and creators needing descriptive and structured prompts for image generation tools.
  • General Conversational AI: As a fast and versatile generalist for everyday questions and companionship in its "non-thinking" mode.

Finetunes
2 models