Sentient AI
Information
Language Models
Language models are capable of understanding the context of a sentence and generating a response that is coherent and relevant.
Sentient Offers variety of free Public Language models including state of the art Sentient-70B-V capable of generating human like responses. While models like Sentient-8x7B focusing on more user friendly responses and ability to interpret emojis while also using them.
There are also guard models such as Sentient-8B-G for users to use in a more controlled environment with its ability to detect and filter out inappropriate content.
There are also guard models such as Sentient-8B-G for users to use in a more controlled environment with its ability to detect and filter out inappropriate content.
Optional Configuration
Optional Configuration are parameters passed onto model in the request to modify the output of the model. The following are the optional configurations available for the language models:
| Parameter | Description | Value |
|---|---|---|
| top_p | Controls nucleus sampling by selecting the smallest set of tokens whose cumulative probability is greater than or equal to the specified p value, ensuring that the model focuses on the most likely next words, reducing randomness. | Float |
| max_tokens | Limits the total number of tokens (words or characters) that the model can generate in a response, ensuring the output stays within a specified length. | Integer |
| stop | Specifies one or more sequences where the model should stop generating text, allowing you to control when the response ends, such as after a certain phrase or word. | String, Array |
| temperature | Adjusts the creativity of the model's responses, with higher values (e.g., 1.0) making output more random, and lower values (e.g., 0.2) making it more focused and deterministic. | Float |
Core Models
1. Sentient-9B-IT
General Information
- Description: Sentient-9B-IT is a 9-billion parameter instruction-tuned AI model from Google's Gemma 2 family. It's optimized for text generation, reasoning, and conversational AI tasks. This model is designed to be lightweight yet powerful, making it suitable for on-device AI applications and cloud-based inference. It performs well in code generation, question answering, and summarization while being efficient in resource usage compared to larger models.
- Version: V1
- Type: Gemma
API Details
- Endpoint: /api/v1/chat/completions
- Parameters:
- Input: Text
- Output: Text
- Optional Configuration: top_p, temperature, stop, max_tokens
Performance and Metrics
- Accuracy: 95%
- Languages: English, Arabic, Russian, Chinese, etc.
- Limitations:
- Does not support images or videos.
- Limited context length.
Additional Information
- Developer: Google
- Status: Production
2. Sentient-70B-V
General Information
- Description: Sentient-70B-V is a 70-billion parameter language model from Meta’s Llama 3 family, featuring an 8,192-token context window for improved long-form reasoning, text generation, and code assistance. It is instruction-tuned, making it highly effective for chatbots, research, and complex problem-solving while being open-weight for customization and fine-tuning.
- Version: V1
- Type: GPT
API Details
- Endpoint: /api/v1/chat/completions
- Parameters:
- Input: Text
- Output: Text
- Optional Configuration: top_p, temperature, stop, max_tokens
Performance and Metrics
- Accuracy: 97%
- Languages: English, Arabic, Russian, Chinese, etc.
- Limitations:
- Does not support images or videos.
- Slow text generation.
Additional Information
- Developer: Meta
- Status: Production
3. Sentient-8B-I
General Information
- Description: Sentient-8B-I is an 8-billion parameter language model from Meta’s Llama 3.1 family, optimized for speed and efficiency while maintaining strong text generation, reasoning, and coding capabilities. It is designed for low-latency applications, making it ideal for chatbots, AI assistants, and real-time processing, while still being open-weight for fine-tuning and deployment flexibility.
- Version: V1
- Type: GPT
API Details
- Endpoint: /api/v1/chat/completions
- Parameters:
- Input: Text
- Output: Text
- Optional Configuration: top_p, temperature, stop, max_tokens
Performance and Metrics
- Accuracy: 93%
- Languages: English, Arabic, Russian, Chinese, etc.
- Limitations:
- Does not support images or videos.
- Small Context window.
Additional Information
- Developer: Meta
- Status: Production
4. Sentient-8B-G (Guard)
General Information
- Description: Sentient-8B-I is an 8-billion parameter language model from Meta's Llama Guard series, which focuses on enhancing safety and reliability in AI-generated responses. Designed to detect and mitigate harmful or biased content, it’s optimized for tasks like moderation, filtering, and safe interactions in conversational AI systems. While it retains the strong capabilities of Llama models in text generation and reasoning, it includes additional safeguards for ethical AI usage.
- Version: V1
- Type: Guard
API Details
- Endpoint: /api/v1/chat/completions
- Parameters:
- Input: Text
- Output: Boolean
- Optional Configuration: None
Performance and Metrics
- Accuracy: 95%
- Languages: English, Arabic, Russian, Chinese, etc.
- Limitations:
- Does not support images or videos.
- Small Context window.
- Limited to Guard functionality.
Additional Information
- Developer: Meta
- Status: Production
5. Sentient-8x7B
General Information
- Description: Sentient-8B-I is a multi-model language system consisting of 8 models, each with 7 billion parameters, designed for scalable, parallelized processing. It features a massive 32,768 token context length, allowing it to handle extremely long input sequences efficiently. Optimized for tasks like text generation, reasoning, and multi-turn conversations, Sentient-8x7B excels in complex applications such as long-form content generation, large document processing, and high-performance AI tasks, offering a blend of high capacity and flexibility.
- Version: V1
- Type: GPT
API Details
- Endpoint: /api/v1/chat/completions
- Parameters:
- Input: Text
- Output: Text
- Optional Configuration: top_p, temperature, stop, max_tokens
Performance and Metrics
- Accuracy: 97%
- Languages: English, Arabic, Russian, Chinese, etc.
- Limitations:
- Does not support images or videos.
Additional Information
- Developer: Mistral
- Status: Production
PreviousModels
On this Page
Language Models
Optional Configuration
1. Sentient-9B-IT
2. Sentient-70B-V
3. Sentient-8B-I
4. Sentient-8B-G
5. Sentient-8x7B