SiliconFlow

SiliconFlow

active

硅基流动 — Chinese inference platform with a permanent free tier for models ≤9B parameters (Qwen3-8B, GLM-4-9B, DeepSeek R1 Distill 7B and more).

https://api.siliconflow.com/v1

Avg Latency

90-day Uptime

%

Rate Limits

1,000 RPM / — RPD

Sign-up Required

Yes

Info

Base URL https://api.siliconflow.com/v1
Sign-up Required Yes
Credit Card Not required
Context Window 131K
Last Verified 2026-05-20
china-availableopenai-compatibledeepseekllamavisionthinking

Models (10)

Qwen/Qwen3-8B

CTX: 131K · 1000 RPM

active

Qwen/Qwen2.5-7B-Instruct

CTX: 32K · 1000 RPM

active

Qwen/Qwen2.5-Coder-7B-Instruct

CTX: 32K · 1000 RPM

active

Qwen/Qwen2.5-VL-7B-Instruct

CTX: 32K · 1000 RPM

active

deepseek-ai/DeepSeek-R1-Distill-Qwen-7B

CTX: 131K · 1000 RPM

active

meta-llama/Meta-Llama-3.1-8B-Instruct

CTX: 32K · 1000 RPM

active

THUDM/GLM-4-9B-0414

CTX: 32K · 1000 RPM

active

THUDM/GLM-Z1-9B-0414

CTX: 131K · 1000 RPM

active

zai-org/GLM-4.1V-9B-Thinking

CTX: 65K · 1000 RPM

active

tencent/Hunyuan-MT-7B

CTX: 32K · 1000 RPM

active

Quick Start

bash
export API_KEY="your_api_key_here"
curl https://api.siliconflow.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $API_KEY" \
  -d '{
    "model": "Qwen/Qwen3-8B",
    "messages": [
      {"role": "user", "content": "Hello! How are you?"}
    ]
  }'