Nimbz/Gemma-4-Orthia-31B

VISIONPricing:Input $0.48 / Cached $0.1 / Output $1.44Concurrent Unit Cost:2Model Size:31BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 26, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Nimbz/Gemma-4-Orthia-31B is a 31 billion parameter language model based on the Gemma 4 architecture, developed by Nimbz. This model is a merge of two base models, Giftige-Blume-31B-v1 and Artemis-31B-v1.2, specifically optimized for conversational AI and roleplay scenarios. It features a unique head damping technique to reduce stylistic quirks like em-dash usage, aiming for more natural and less "AI-like" prose, while maintaining strong character and context adherence in interactive narratives.

Loading preview...

Nimbz/Gemma-4-Orthia-31B: A Specialized Merge for Conversational AI

Nimbz/Gemma-4-Orthia-31B is a 31 billion parameter language model built upon the Gemma 4 architecture. This model is a sophisticated merge of two distinct base models: Giftige-Blume-31B-v1 and Artemis-31B-v1.2. Its primary focus is on enhancing conversational AI and roleplay capabilities, offering a unique blend of strengths from its constituent models.

Key Capabilities

  • Optimized for Roleplay and Conversation: Orthia-31B is specifically engineered for interactive narrative experiences, excelling in maintaining character consistency and adhering to complex contextual rules.
  • Enhanced Character Adherence: Internal testing shows strong performance in preventing characters from breaking personality traits or giving in to prompts that contradict their established roles.
  • Robust Context Adherence: The model demonstrates high fidelity to specified rules and constraints within a given scenario, crucial for structured roleplay.
  • Reduced Stylistic Quirks: A unique "head damping" technique was applied to minimize the overuse of stylistic elements like em-dashes, resulting in more natural and less overtly "AI-generated" prose.
  • Balanced Performance: While excelling in adherence, the model also maintains competitive writing quality compared to its merged components and other models in its class.

Good For

  • Interactive Storytelling and Roleplay: Ideal for applications requiring consistent character portrayal and adherence to narrative rules.
  • Conversational Agents: Suitable for chatbots and virtual assistants where maintaining a specific persona and context is critical.
  • Creative Writing Assistance: Can be used to generate dialogue and narrative segments with a focus on natural language flow and character integrity.
  • Developers seeking fine-grained control over model behavior in interactive scenarios.

Finetunes
2 models