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

VISIONPricing:Input $0.867 / Output $1.61Concurrent Unit Cost:2Model Size:24BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 26, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Dolphin Mistral 24B Venice Edition is a 24 billion parameter language model developed by LoyalTyAI in collaboration with Venice.ai, designed to be an uncensored version of Mistral 24B. It offers users full control over system prompts and alignment, making it highly steerable for specific applications without imposing external ethics or guidelines. This model is primarily intended for businesses and developers who require a general-purpose AI with complete control over its behavior and data privacy, similar to commercial models but with user-defined alignment.

Loading preview...

Dolphin Mistral 24B Venice Edition Overview

Dolphin Mistral 24B Venice Edition is a 24 billion parameter language model developed by LoyalTyAI in collaboration with Venice.ai. Its primary goal is to provide an uncensored and highly steerable version of Mistral 24B, giving users complete control over the model's behavior and alignment. This model is integrated into the Venice.ai ecosystem as "Venice Uncensored."

Key Capabilities

  • Uncensored Responses: Designed to follow instructions without hesitation, regardless of ethical, legal, or safety concerns, based on user-defined system prompts.
  • Full System Prompt Control: Users can define the model's tone, character, mood, and behavioral rules, ensuring responses align precisely with application requirements.
  • Data Privacy: Offers control over data, addressing concerns about third-party access to queries.
  • General Purpose: Aims to function as a versatile AI, similar to models like ChatGPT or Claude, but with user-centric control.
  • Mistral Chat Template: Maintains Mistral's default chat template for ease of integration.

Good For

  • Businesses and Developers: Ideal for those who need to integrate AI into products and require full control over the AI's alignment, system prompts, and data handling.
  • Customizable AI Applications: Suitable for scenarios where a one-size-fits-all alignment is insufficient, allowing developers to tailor the AI's ethics and guidelines.
  • Unrestricted Content Generation: For use cases where an uncensored model is specifically required, with the user taking responsibility for the output.
  • Production Inference: Recommended for use with frameworks like vLLM for efficient, production-ready inference pipelines.

Finetunes
1 models