DeepSeek

Overview

DeepSeek builds cost-effective, OpenAI-compatible models with strong mathematical and coding performance, including reasoning models that expose their thinking process.

Official Website: https://www.deepseek.com API Documentation: https://api-docs.deepseek.com

Key features

• Cost-effective — competitive pricing for high-volume workloads
• Reasoning — reasoning models (e.g. deepseek-reasoner / R1) expose reasoning_content
• Strong math & coding — excellent at mathematical reasoning and code generation
• OpenAI-compatible — drop-in chat completions with function calling
• Context caching — automatic caching of repeated prompt prefixes

Using DeepSeek with Yunxin

Address DeepSeek models by prefixing the model id with the deepseek slug — deepseek/<model> (for example deepseek/deepseek-reasoner). You always call the same Yunxin base URL, https://api.yuhuanstudio.com/v1, with your sk- API key; Yunxin routes the request to DeepSeek.

Usage example

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["YUNXIN_API_KEY"],
    base_url="https://api.yuhuanstudio.com/v1",
)

response = client.chat.completions.create(
    model="deepseek/model-id",
    messages=[
        {"role": "user", "content": "Write a Python quicksort."},
    ],
)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.YUNXIN_API_KEY,
  baseURL: "https://api.yuhuanstudio.com/v1",
});

const response = await client.chat.completions.create({
  model: "deepseek/model-id",
  messages: [{ role: "user", content: "Write a Python quicksort." }],
});

console.log(response.choices[0].message.content);

curl https://api.yuhuanstudio.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YUNXIN_API_KEY" \
  -d '{
    "model": "deepseek/model-id",
    "messages": [{"role": "user", "content": "Write a Python quicksort."}]
  }'

Reasoning models

Reasoning models such as deepseek-reasoner (R1) return their thinking in a reasoning_content field alongside the final answer:

{
  "choices": [{
    "message": {
      "role": "assistant",
      "content": "The answer is 42.",
      "reasoning_content": "Let me think through this step by step..."
    }
  }]
}

See Reasoning for how to request and read thinking across formats.

Available models

List the models currently available for this provider:

curl "https://api.yuhuanstudio.com/v1/models?provider=deepseek" \
  -H "Authorization: Bearer $YUNXIN_API_KEY"

Official resources

• DeepSeek Website
• API Documentation
• GitHub