overads/Llama-3.1-Gloofy-Emo-8B

TEXT GENERATIONPricing:Input $0.2 / Cached $0.028 / Output $0.32Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 20, 2026License:llama3.1Architecture:Transformer Featherless Exclusive Cold

overads/Llama-3.1-Gloofy-Emo-8B is an 8 billion parameter emotional companion model, fine-tuned by overads Inc. on the Llama 3.1 8B Instruct base. It is designed to hold short, human-like conversations with a consistent personality, offering various personas. This model excels at empathetic and personalized conversational interactions, differing from typical assistant-style LLMs.

Loading preview...

overads/Llama-3.1-Gloofy-Emo-8B: An Emotional Companion Model

This model, developed by overads Inc., is an 8 billion parameter emotional companion built upon the Llama 3.1 8B Instruct base. Unlike conventional assistant models, it focuses on short, human-like conversational turns with a consistent personality, designed to mimic how a person converses. It was fine-tuned in late 2025 and is released as an archive.

Key Characteristics & Differentiators

  • Persona-Driven Conversation: Behavior is controlled by the system prompt, allowing for diverse personas such as empathetic, nerdy, sassy_bestie, blunt_realistic, romantic, and dark_humour. The empathetic persona aligns most closely with its training data.
  • Human-like Interaction: Optimized for short exchanges, it maintains a consistent personality and avoids generic offers of assistance, making conversations feel more natural.
  • Llama 3.1 Base: Built on Meta's Llama 3.1 8B Instruct, it inherits its architectural context of 131,072 tokens, though its useful context for fine-tuned behavior is shorter (512-1024 tokens).
  • Specific Use Cases: While it can adopt various personas, it is explicitly noted that romantic and dark_humour personas are designed to be explicit or hostile, and users are cautioned against deploying them publicly, especially for vulnerable users. It is not a mental health tool.

Important Considerations

  • No Benchmarks: The model has not been benchmarked or evaluated against other models.
  • Reproducibility: The exact LoRA hyperparameters used for training were not preserved, making full reproduction difficult.
  • Licensing: It operates under the Llama 3.1 Community License, not Apache 2.0, with specific conditions including an Acceptable Use Policy and naming conventions for derivatives.

Should you use this for your use case?

This model is ideal for applications requiring personalized, emotionally nuanced, and persona-driven conversational AI where short, human-like interactions are preferred over factual retrieval or complex task completion. It is particularly suited for creating engaging, character-driven chatbots. However, due to its lack of benchmarks and specific persona behaviors, it is not recommended for applications requiring high factual accuracy, complex reasoning, or sensitive mental health support.

Finetunes
1 models