servantofares/Dolphin-Mistral-24B-Venice-Edition

Hugging Face
TEXT GENERATIONConcurrency Cost:2Model Size:24BQuant:FP8Ctx Length:32kPublished:Mar 21, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

Dolphin-Mistral-24B-Venice-Edition is a 24 billion parameter Mistral-based language model developed collaboratively by servantofares and Venice.ai. This model is specifically fine-tuned for uncensored responses, offering users full control over system prompts and alignment. It aims to be a general-purpose, steerable AI tool, providing an alternative to highly aligned commercial models by allowing users to define ethical and behavioral guidelines. It is designed for applications requiring flexible and user-controlled content generation.

Loading preview...

Dolphin Mistral 24B Venice Edition Overview

Dolphin Mistral 24B Venice Edition is a 24 billion parameter language model, a collaborative effort between servantofares and Venice.ai. It is specifically designed to be an uncensored version of Mistral 24B, providing users with maximum control over its behavior and outputs. This model is the default for Venice.ai users, branded as "Venice Uncensored."

Key Differentiators & Capabilities

  • Uncensored & Steerable: Unlike many commercial models, Dolphin Mistral 24B Venice Edition does not impose its own ethics or guidelines. Users define the system prompt to set the tone, alignment, and rules for its responses.
  • User Control: It addresses common issues with commercial LLMs by giving users control over system prompts, model versions, alignment, and data privacy. The model is presented as a tool where the user is responsible for the generated content.
  • General Purpose: While emphasizing uncensored capabilities, it aims to function as a general-purpose model, similar to those behind ChatGPT, Claude, and Gemini, but with user-defined constraints.
  • Mistral Chat Template: It maintains Mistral's default chat template, ensuring compatibility and ease of use for those familiar with the Mistral ecosystem.

Usage Recommendations

  • System Prompt: Users are strongly encouraged to set a system prompt to define the model's tone and guidelines, as its default behavior is highly flexible.
  • Low Temperature: A relatively low temperature, such as temperature=0.15, is recommended for optimal output quality.
  • Framework Support: The model is compatible with popular frameworks like vLLM (recommended for production inference) and transformers.

Finetunes
1 models