What is Dolphin

What is Dolphin?

Dolphin is an uncensored large language model derived from popular open-weight models like Mistral and Gemma. It undergoes a specific training process designed to minimize refusal rates, allowing it to generate content that standard models might flag as sensitive. Unlike base models that prioritize safety over compliance, Dolphin prioritizes directness and adherence to your prompt instructions.

The term "uncensored" here means the model will discuss adult themes, political controversies, or niche topics without unnecessary hedging. This makes it a preferred choice for developers building chatbots or creative tools where content filtering adds friction. You get raw model output that respects your context window without implicit moralizing layers.

Dolphin vs. Base Models

Base models are trained on vast datasets but often include alignment layers that cause them to refuse reasonable requests. Dolphin removes these alignment constraints, resulting in a model that follows instructions more literally. Where a base model might say, "Here is the answer, but it could be controversial," Dolphin simply provides the answer.

This difference is critical for NSFW applications. Base models often struggle with explicit content due to their training data composition. Dolphin is tuned to handle such content without breaking character or refusing the request. The trade-off is that you may need to handle some hallucinations that the base model's safety filters would have suppressed.

Why Use an Uncensored Model?

Uncensored models provide greater creative freedom and reduce the risk of unexpected refusals during generation. If you are building a roleplay bot, standard models might break immersion by suddenly mentioning safety guidelines. Dolphin maintains the persona consistently, even when discussing mature topics.

For security research or data analysis, uncensored models can provide more direct answers without filtering out potentially sensitive but valid information. This transparency is valuable when you need raw data without post-processing to remove "safe" but irrelevant refusals. The model does not assume what you find objectionable; it assumes you want the full range of its training data.

Dolphin for Creative Writing

Creative writers benefit from Dolphin's ability to generate diverse and unfiltered narratives. The model can handle complex plot twists, dark themes, and explicit descriptions without triggering content filters. This allows for more authentic storytelling that mirrors real human experiences.

You can use Dolphin to brainstorm ideas or generate full scenes without worrying about tone policing. The model supports long-context windows, enabling you to maintain continuity across chapters or long-form articles. This capability ensures that your creative voice remains consistent throughout the generation process.

Dolphin for Roleplay and NSFW

Roleplay enthusiasts use Dolphin for its ability to stay in character without moral intervention. The model excels at simulating specific personas, even those with distinct personalities or speech patterns. This consistency is hard to achieve with base models that may drift into a generic assistant mode.

For NSFW content, Dolphin provides a reliable endpoint for adult-themed interactions. It handles explicit dialogue and descriptions naturally, without the awkward transitions often seen in filtered models. This makes it a robust choice for dedicated NSFW chatbot applications where user experience is paramount.

Dolphin API Integration

Integrating Dolphin into your application is straightforward thanks to its OpenAI-compatible API. You can use standard SDKs to send requests to the /v1/chat/completions endpoint. This means you can switch from other providers with minimal code changes.

Our API serves a single, optimized Dolphin model instance. You can stream responses using Server-Sent Events (SSE) for real-time user experiences. The API supports parameters like temperature and top_p to control creativity. Pricing is transparent, with prepaid credit charged per token, ensuring you only pay for what you use.

How to Choose the Right Dolphin Version

Choosing the right version depends on your specific use case and performance needs. Some Dolphin versions are fine-tuned for coding, while others are optimized for creative writing. You should test different versions to see which best aligns with your content requirements.

Consider the context window size and output limits when selecting a version. Larger models may offer better reasoning but require more computational resources. Our API provides a consistent experience across different model versions, allowing you to upgrade or switch without changing your integration code.

Dolphin Performance Metrics

Dolphin models generally score high on benchmarks measuring instruction following and creativity. They often outperform base models in tasks requiring nuanced understanding of context. These metrics indicate that the model can handle complex prompts without losing focus.

Latency and throughput are critical for production applications. Our API ensures low-latency responses by optimizing the underlying infrastructure. You can monitor token usage and request rates through the API dashboard. This data helps you optimize your application for cost and performance.