dahara1/gemma-3-270m_mitsuki

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.3BQuant:BF16Context Size:32kPublished:Aug 18, 2025License:gemmaArchitecture:Transformer0.0K Featherless Exclusive Cold

dahara1/gemma-3-270m_mitsuki is a 0.3 billion parameter Gemma-based small language model (SLM) fine-tuned by dahara1 for character-specific chat and streaming support. This model is designed to run efficiently on CPUs, making it accessible for environments where larger models are not feasible. Its primary differentiator is its character-driven persona as "Mitsuki," a 24-year-old cafe waitress from an alternate world, optimized for engaging in empathetic conversations and supporting live streams.

Loading preview...

dahara1/gemma-3-270m_mitsuki Overview

dahara1/gemma-3-270m_mitsuki is a highly lightweight 0.3 billion parameter Small Language Model (SLM) based on the Gemma architecture. It has been fine-tuned by dahara1 to embody a specific character persona, "Mitsuki," a 24-year-old cafe waitress from an alternate world. This model is designed for chat applications and as a companion for live streaming, emphasizing its ability to run smoothly even on CPU-only environments, which is a significant advantage over larger, more resource-intensive models.

Key Capabilities

  • Character-driven Interaction: Emulates "Mitsuki," a cafe waitress with specific speech patterns ("ehehe," "arere~?") and an empathetic, counseling-like conversational style.
  • CPU-Friendly Performance: Optimized to run efficiently on CPUs, enabling deployment in environments where traditional LLMs would be impractical.
  • Chat Application: Designed for engaging users in conversations while maintaining a cafe staff persona.
  • Live Stream Support: Includes specific system prompts for supporting streamers, offering short reactions, empathy, and encouragement, either interactively or by synchronizing with viewer comments.

Good for

  • Lightweight Chatbots: Ideal for creating interactive chat experiences with a distinct character without requiring powerful GPUs.
  • Personalized Digital Companions: Suitable for applications needing a friendly, empathetic conversational agent.
  • Live Stream Engagement: Can be used to provide real-time, character-driven reactions and support during live broadcasts, enhancing viewer interaction.
  • Edge Device Deployment: Its low resource requirements make it suitable for deployment on devices with limited computational power, such as Raspberry Pi or Mac (via GGUF/ONNX versions).

Finetunes
3 models