Kimi-K2 on SiliconFlow: Tailored for AI Agents, Priced to Scale

MoonShot AI's powerful open-source Kimi K2 Mixture-of-Experts (MoE) model is now available on SiliconFlow. Developers can seamlessly integrate this model – featuring 32B activated parameters and 1 trillion total parameters – via SiliconFlow's production-ready API to build advanced coding tools and agentic applications.

SiliconFlow supports:

• High-Speed Inference: Optimized for lower latency and higher throughput.

• Cost-Optimized Pricing: $0.58/M tokens (input) and $2.29/M tokens (output).

• Extended Context Window: 128K context window for complex tasks.

• TPM Quota: 100,000 tokens per minute.

Key Technical Highlights of Kimi K2:

• Large-Scale Training: Pre-trained a 1T-parameter MoE model on 15.5T tokens with zero training instability.

• MuonClip Optimizer: Apply the Muon optimizer to an unprecedented scale, and develop novel optimization techniques to resolve instabilities while scaling up.

• Agentic Intelligence: Tailored for AI agents -- tool use, reasoning, and autonomous problem-solving.

Get Started Immediately

Explore: Try Kimi-K2-Instruct in the SiliconFlow playground.Integrate: Use our OpenAI-compatible API. Explore the full API specifications in the SiliconFlow API documentation.

from openai import OpenAI

url = 'https://api.ap.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": "Explain the concept of gravitational waves in Chinese?"}
]
response = client.chat.completions.create(
    model="moonshotai/Kimi-K2-Instruct",
    messages=messages,
    stream=True,  # Enable streaming output
    max_tokens=8192
)
# Gradually receive and process the response
for chunk in response:
    if chunk.choices[0].delta.content:
        content += chunk.choices[0].delta.content

Kimi K2’s exceptional capabilities in code generation and agentic reasoning make it a powerful tool for developers across various domains. For coding, it automates repetitive tasks, generates optimized code, and assists in debugging and refactoring. In agent-related tasks, it can analyze data, plan complex itineraries, and automate workflows by orchestrating multiple tools.

Build with the Kimi-K2-Instruct API on SiliconFlow today!