win10/Muse-Glimmer-30B-EXP-EVE-v1

VISIONPricing:Input $1.2 / Cached $0.04 / Output $4.4Concurrent Unit Cost:2Model Size:30BQuant:FP8Context Size:128kPublished:Aug 23, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

win10/Muse-Glimmer-30B-EXP-EVE-v1 is a 30 billion parameter cross-architecture language model, developed by win10, built upon the Muse-Glimmer-30B backbone. Utilizing a Tensor Gene Evolution merging approach, it integrates capabilities from DeepSeek-V4-Flash-0731 and dots3-note-prev. This model is optimized for enhanced reasoning, demonstrating broader coverage, greater depth of thought, and more diverse problem-solving behaviors compared to its base model.

Loading preview...

Muse-Glimmer-30B-EXP-EVE-v1 Overview

win10/Muse-Glimmer-30B-EXP-EVE-v1 is a 30 billion parameter language model that represents a significant advancement in model merging techniques. Developed by win10, this model is a cross-architecture merge built around the robust Muse-Glimmer-30B as its primary foundation.

Key Capabilities and Merging Approach

This model leverages a Tensor Gene Evolution merging strategy to incorporate distinct capabilities and behavioral characteristics from two prominent donor models:

  • deepseek-ai/DeepSeek-V4-Flash-0731
  • dots-studio/dots3-note-prev

Enhanced Reasoning and Problem-Solving

The primary differentiator of Muse-Glimmer-30B-EXP-EVE-v1 lies in its significantly improved reasoning characteristics compared to the original Muse-Glimmer base model. Users can expect:

  • Broader reasoning coverage
  • Greater depth of thought
  • More diverse problem-solving behavior

This makes the model particularly well-suited for tasks requiring complex logical inference and varied analytical approaches, while preserving the strengths of its donor models.

Finetunes
1 models