DeepSeek-V3-0324 Is Now Live on SiliconFlow

Mar 27, 2025

DeepSeek-V3-0324 (671B) is now live on SiliconFlow — delivering major improvements in reasoning, writing, and math performance.

Reinforcement learning boosts performance on complex reasoning tasks
8K token context window
Competitive price: $0.29/M tokens (input), $1.15/M tokens (output)

DeepSeek-V3-0324 keeps the same model size, pricing, and API compatibility — but brings major gains in reasoning, writing, and math. With support for Function Calling, JSON mode, Prefix, FIM, it’s a small version that delivers a big upgrade.

Quick Start

Try the DeepSeek-V3-0324 on the SiliconFlow playground.

Quick Access to API

The following Python example demonstrates how to call the DeepSeek-V3-0324 model via SiliconFlow’s API endpoint. More detailed API specifications for developers.

from openai import OpenAI

url = 'https://api.siliconflow.com/v1/'
api_key = 'your api_key'

client = OpenAI(
    base_url=url,
    api_key=api_key
)

# Send a request with streaming output
content = ""
reasoning_content = ""
messages = [
    {"role": "user", "content": "Prove the Pythagorean theorem and provide a simple example."}
]
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True,  # Enable streaming output
    max_tokens=4096,
    extra_body={
        "thinking_budget": 1024
    }
)
# Gradually receive and process the response
for chunk in response:
    if chunk.choices[0].delta.content:
        content += chunk.choices[0].delta.content
    if chunk.choices[0].delta.reasoning_content:
        reasoning_content += chunk.choices[0].delta.reasoning_content

# Round 2
messages.append({"role": "assistant", "content": content})
messages.append({'role': 'user', 'content': "Continue"})
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True
)

from openai import OpenAI

url = 'https://api.siliconflow.com/v1/'
api_key = 'your api_key'

client = OpenAI(
    base_url=url,
    api_key=api_key
)

# Send a request with streaming output
content = ""
reasoning_content = ""
messages = [
    {"role": "user", "content": "Prove the Pythagorean theorem and provide a simple example."}
]
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True,  # Enable streaming output
    max_tokens=4096,
    extra_body={
        "thinking_budget": 1024
    }
)
# Gradually receive and process the response
for chunk in response:
    if chunk.choices[0].delta.content:
        content += chunk.choices[0].delta.content
    if chunk.choices[0].delta.reasoning_content:
        reasoning_content += chunk.choices[0].delta.reasoning_content

# Round 2
messages.append({"role": "assistant", "content": content})
messages.append({'role': 'user', 'content': "Continue"})
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True
)

from openai import OpenAI

url = 'https://api.siliconflow.com/v1/'
api_key = 'your api_key'

client = OpenAI(
    base_url=url,
    api_key=api_key
)

# Send a request with streaming output
content = ""
reasoning_content = ""
messages = [
    {"role": "user", "content": "Prove the Pythagorean theorem and provide a simple example."}
]
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True,  # Enable streaming output
    max_tokens=4096,
    extra_body={
        "thinking_budget": 1024
    }
)
# Gradually receive and process the response
for chunk in response:
    if chunk.choices[0].delta.content:
        content += chunk.choices[0].delta.content
    if chunk.choices[0].delta.reasoning_content:
        reasoning_content += chunk.choices[0].delta.reasoning_content

# Round 2
messages.append({"role": "assistant", "content": content})
messages.append({'role': 'user', 'content': "Continue"})
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=messages,
    stream=True
)

Optimized for demanding reasoning and real-world use, DeepSeek-V3-0324 is ready to help you build smarter tools, better agents, and long-form assistants.

Try DeepSeek-V3 on SiliconFlow now and see what you can create.