Vortex5/Glimmering-Citrus-31B

VISIONConcurrent Unit Cost:2Model Size:31BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 18, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Vortex5/Glimmering-Citrus-31B is a 31 billion parameter language model created by merging Equinox-31B, Gemma4-Garnet-31B, G4-MeroMero-v2-31B, and Gemma-4-31B-it using a custom ARCS method. This model is specifically designed and optimized for creative applications such as roleplay, creative writing, storytelling, and brainstorming. It leverages a 32768 token context length to support detailed and extended generative tasks.

Loading preview...

Glimmering-Citrus-31B: A Merged Model for Creative Generation

Glimmering-Citrus-31B is a 31 billion parameter language model developed by Vortex5. It was constructed through a unique merge process, combining four distinct base models: LatitudeGames/Equinox-31B, ConicCat/Gemma4-Garnet-31B, zerofata/G4-MeroMero-v2-31B, and google/gemma-4-31B-it. This merge was performed using a custom ARCS method, with specific parameters for strength, sharp, voice, and envelope, aiming to synthesize the strengths of its constituent models.

Key Capabilities

  • Roleplay: Excels in character interaction, dialogue generation, persona adoption, and multi-turn scene development.
  • Creative Writing: Proficient in drafting fiction, generating detailed descriptions, crafting dialogue, and exploring scene variations.
  • Storytelling: Supports the creation of short stories, longer narratives, plot development, and worldbuilding.
  • Brainstorming: Effective for generating concepts, outlines, alternative ideas, and various writing prompts.

Technical Details

The model utilizes google/gemma-4-31B-it as its base and processes data with bfloat16 output dtype. Its tokenizer is sourced from a union of the merged models, ensuring broad compatibility. With a context length of 32768 tokens, Glimmering-Citrus-31B is well-suited for tasks requiring extensive contextual understanding and generation.

Intended Use Cases

This model is specifically tailored for generative tasks that demand creativity and nuanced language understanding. Developers and users looking for a model optimized for imaginative content creation, interactive narratives, and idea generation will find Glimmering-Citrus-31B particularly useful.