Enterprise Plans
AI Application Development
AI Translator API
AI SEO/GEO Service
Web Scraping Service
Log in
const { OpenAI } = require('openai');
const api = new OpenAI({
baseURL: 'https://api.ai.cc/v1',
apiKey: '',
});
const main = async () => {
const result = await api.chat.completions.create({
model: 'deepseek/deepseek-r1',
messages: [
{
role: 'system',
content: 'You are an AI assistant who knows everything.',
},
{
role: 'user',
content: 'Tell me, why is the sky blue?'
}
],
});
const message = result.choices[0].message.content;
console.log(`Assistant: ${message}`);
};
main();
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.ai.cc/v1",
api_key="",
)
response = client.chat.completions.create(
model="deepseek/deepseek-r1",
messages=[
{
"role": "system",
"content": "You are an AI assistant who knows everything.",
},
{
"role": "user",
"content": "Tell me, why is the sky blue?"
},
],
)
message = response.choices[0].message.content
print(f"Assistant: {message}")
-
AI Playground
Test all API models in the sandbox environment before you integrate.
We provide more than 300 models to integrate into your app.
Product Detail
DeepSeek R1: The Next Generation AI Reasoning Model
DeepSeek R1, developed by DeepSeek AI, represents a significant leap in Large Language Model (LLM) technology. Launched on January 21, 2025, this version 1.0 model is meticulously engineered to excel in complex problem-solving, advanced mathematical reasoning, and comprehensive programming assistance.
✨ Core Details:
- Model Name: DeepSeek R1
- Developer/Creator: DeepSeek AI
- Release Date: January 21, 2025
- Version: 1.0
- Model Type: Large Language Model (LLM) with a strong focus on reasoning capabilities.
🚀 Unrivaled Features & Performance
DeepSeek R1 stands out with its innovative architecture and training approach, offering exceptional performance for a wide array of demanding applications.
- 🧠 Mixture-of-Experts (MoE) Architecture: Optimizes computational resources by activating only 37 billion out of its 671 billion parameters per token. This intelligent design ensures efficiency without sacrificing high performance.
- 💡 Advanced Chain-of-Thought Reasoning: Empowers the model to dissect complex problems into smaller, manageable steps, significantly boosting the clarity and accuracy of its problem-solving.
- 📈 Superior Benchmark Performance: Achieves impressive scores across various benchmarks, including a remarkable 91.6% on the MATH benchmark and competitive results in coding challenges.
- 🧪 Pure Reinforcement Learning Training: Utilizes a unique training methodology based on pure reinforcement learning, bypassing extensive supervised fine-tuning to organically enhance its core reasoning capabilities.
- 📜 Open-Source MIT License: Available under the MIT license, enabling unrestricted use and modification for both research and commercial endeavors.
💰 DeepSeek R1 Pricing Structure
Input Cost: $0.0006064 per 1000 tokens
Output Cost: $0.0024145 per 1000 tokens
Please note: These special introductory prices are valid until February 15. Prices will be adjusted thereafter.
🎯 Who Can Benefit from DeepSeek R1?
DeepSeek R1 is engineered for professionals requiring advanced reasoning capabilities, making it an invaluable asset for software developers, data scientists, and researchers. Key applications include:
- Mathematical Computations: Solving intricate equations and performing complex calculations.
- Coding Challenges: Assisting with diverse programming tasks, including debugging and efficient code generation.
- Logical Problem-Solving: Tackling complex logical puzzles and enabling informed decision-making.
While DeepSeek R1 primarily supports English, it offers adaptability for multiple languages based on specific user requirements.
⚙️ Technical Deep Dive
Architecture
DeepSeek R1's foundation is its advanced Mixture-of-Experts (MoE) architecture. This intelligent design ensures that only a fractional portion of its vast parameter count is activated during each forward pass, significantly lowering computational costs while consistently maintaining peak performance levels across various tasks.
Training Data
The model was trained on an expansive and highly diverse dataset to ensure robust and versatile performance.
- Data Source & Size: The training dataset encompasses a staggering 14.8 trillion tokens, meticulously curated from a broad spectrum of publicly available code repositories and comprehensive mathematical texts.
- Diversity & Bias: Extensive efforts were made during dataset curation to minimize biases and maximize diversity in topics and styles, guaranteeing highly reliable and adaptable performance across different scenarios.
Performance Metrics
A visual overview illustrating DeepSeek R1's robust performance across key benchmarks:
💻 Integrating DeepSeek R1 into Your Projects
DeepSeek R1 is conveniently accessible on the AI/ML API platform. You can identify and utilize the model by its designated name: "DeepSeek R1".
Code Samples
A typical way to reference DeepSeek R1 via API is as follows:
// Example snippet for API integration <snippet data-name="open-ai.chat-completion" data-model="deepseek/deepseek-r1"></snippet>
Comprehensive API Documentation
For detailed guidance on integration, available endpoints, and best practices, please consult the official DeepSeek R1 API Documentation.
📜 Ethical Framework & Licensing
Ethical Guidelines
DeepSeek AI is deeply committed to ethical AI development, advocating for transparency regarding DeepSeek R1's inherent capabilities and limitations. Users are strongly encouraged to practice responsible usage to mitigate any potential misuse or harmful applications of content generated by the model.
Licensing
DeepSeek R1 is distributed under an open-source MIT license. This permissive license grants extensive usage rights for both academic research and commercial projects, ensuring compliance with ethical standards and creator rights.
Get DeepSeek R1 API Access Today!
❓ Frequently Asked Questions (FAQ)
Q: What is DeepSeek R1 and what are its primary capabilities?
A: DeepSeek R1 is a cutting-edge Large Language Model (LLM) from DeepSeek AI, specialized in advanced reasoning, complex mathematical problem-solving, and robust programming assistance. It utilizes a highly efficient Mixture-of-Experts (MoE) architecture.
Q: What makes DeepSeek R1 stand out from other LLMs?
A: Its unique Mixture-of-Experts (MoE) architecture allows for efficient parameter activation, and its training via pure reinforcement learning (without extensive supervised fine-tuning) contributes to its exceptional performance on reasoning benchmarks, such as 91.6% on MATH.
Q: Who is the ideal user for DeepSeek R1?
A: DeepSeek R1 is perfectly suited for software developers, data scientists, and researchers who require sophisticated AI capabilities for mathematical computations, resolving coding challenges, and intricate logical problem-solving within their applications.
Q: What are the licensing terms for DeepSeek R1?
A: DeepSeek R1 is released under an open-source MIT license, which permits both research and commercial usage. This ensures flexibility for developers while upholding ethical guidelines and creator rights.
Q: How can I access DeepSeek R1 for my projects?
A: You can gain access to DeepSeek R1 through the AI/ML API platform. For comprehensive technical details and integration instructions, refer to the dedicated API Documentation.
Learn how you can transformyour company with AICC APIs
