跳至主要內容

SiliconFlow （中國） 標誌SiliconFlow （中國）

透過 Mastra 的模型路由器存取 47 個 SiliconFlow （中國） 模型。系統會自動使用 SILICONFLOW_CN_API_KEY 環境變數進行驗證。

詳情請參閱SiliconFlow （中國） 文件。

.env
SILICONFLOW_CN_API_KEY=your-api-key
src/mastra/agents/my-agent.ts
import { Agent } from '@mastra/core/agent'

const agent = new Agent({
  id: 'my-agent',
  name: 'My Agent',
  instructions: 'You are a helpful assistant',
  model: 'siliconflow-cn/ByteDance-Seed/Seed-OSS-36B-Instruct',
})

// Generate a response
const response = await agent.generate('Hello!')

// Stream a response
const stream = await agent.stream('Tell me a story')
for await (const chunk of stream) {
  console.log(chunk)
}
資訊

Mastra 使用與 OpenAI 相容的 /chat/completions 端點。部分 Provider 專屬功能可能無法使用，詳情請參閱SiliconFlow （中國） 文件。

模型
模型 的直接連結

ModelContextToolsReasoningImageAudioVideoInput $/1MOutput $/1M
siliconflow-cn/baidu/ERNIE-4.5-300B-A47B131K$0.28$1
siliconflow-cn/ByteDance-Seed/Seed-OSS-36B-Instruct262K$0.21$0.57
siliconflow-cn/deepseek-ai/DeepSeek-OCR8K——
siliconflow-cn/deepseek-ai/DeepSeek-R1164K$0.50$2
siliconflow-cn/deepseek-ai/DeepSeek-V3164K$0.25$1
siliconflow-cn/deepseek-ai/DeepSeek-V3.1-Terminus164K$0.27$1
siliconflow-cn/deepseek-ai/DeepSeek-V3.2164K$0.27$0.42
siliconflow-cn/deepseek-ai/DeepSeek-V4-Flash1.0M$0.14$0.28
siliconflow-cn/deepseek-ai/DeepSeek-V4-Pro1.0M$2$3
siliconflow-cn/inclusionAI/Ling-flash-2.0131K$0.14$0.57
siliconflow-cn/PaddlePaddle/PaddleOCR-VL-1.516K——
siliconflow-cn/Pro/deepseek-ai/DeepSeek-R1164K$0.50$2
siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3164K$0.25$1
siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3.1-Terminus164K$0.27$1
siliconflow-cn/Pro/deepseek-ai/DeepSeek-V3.2164K$0.27$0.42
siliconflow-cn/Pro/MiniMaxAI/MiniMax-M2.5192K$0.30$1
siliconflow-cn/Pro/moonshotai/Kimi-K2.5262K$0.45$2
siliconflow-cn/Pro/moonshotai/Kimi-K2.6262K$0.95$4
siliconflow-cn/Pro/zai-org/GLM-5205K$1$3
siliconflow-cn/Pro/zai-org/GLM-5.1205K$1$4
siliconflow-cn/Qwen/Qwen2.5-72B-Instruct33K$0.59$0.59
siliconflow-cn/Qwen/Qwen2.5-7B-Instruct33K$0.05$0.05
siliconflow-cn/Qwen/Qwen3-14B131K$0.07$0.28
siliconflow-cn/Qwen/Qwen3-235B-A22B-Thinking-2507262K$0.13$0.60
siliconflow-cn/Qwen/Qwen3-30B-A3B-Instruct-2507262K$0.09$0.30
siliconflow-cn/Qwen/Qwen3-32B131K$0.14$0.57
siliconflow-cn/Qwen/Qwen3-8B131K$0.06$0.06
siliconflow-cn/Qwen/Qwen3-Coder-30B-A3B-Instruct262K$0.07$0.28
siliconflow-cn/Qwen/Qwen3-Coder-480B-A35B-Instruct262K$0.25$1
siliconflow-cn/Qwen/Qwen3-VL-235B-A22B-Instruct262K$0.30$2
siliconflow-cn/Qwen/Qwen3-VL-235B-A22B-Thinking262K$0.45$4
siliconflow-cn/Qwen/Qwen3-VL-30B-A3B-Instruct262K$0.29$1
siliconflow-cn/Qwen/Qwen3-VL-30B-A3B-Thinking262K$0.29$1
siliconflow-cn/Qwen/Qwen3-VL-32B-Instruct262K$0.20$0.60
siliconflow-cn/Qwen/Qwen3-VL-32B-Thinking262K$0.20$2
siliconflow-cn/Qwen/Qwen3-VL-8B-Instruct262K$0.18$0.68
siliconflow-cn/Qwen/Qwen3.5-122B-A10B262K$0.29$2
siliconflow-cn/Qwen/Qwen3.5-27B262K$0.26$2
siliconflow-cn/Qwen/Qwen3.5-35B-A3B262K$0.23$2
siliconflow-cn/Qwen/Qwen3.5-397B-A17B262K$0.29$2
siliconflow-cn/Qwen/Qwen3.5-4B262K——
siliconflow-cn/Qwen/Qwen3.5-9B262K$0.22$2
siliconflow-cn/Qwen/Qwen3.6-35B-A3B262K$0.23$2
siliconflow-cn/stepfun-ai/Step-3.5-Flash262K$0.10$0.30
siliconflow-cn/tencent/Hunyuan-A13B-Instruct131K$0.14$0.57
siliconflow-cn/zai-org/GLM-4.5-Air131K$0.14$0.86
siliconflow-cn/zai-org/GLM-5.21.0M$1$4
47 available models

進階設定
進階設定 的直接連結

自訂標頭
自訂標頭 的直接連結

src/mastra/agents/my-agent.ts
const agent = new Agent({
  id: 'custom-agent',
  name: 'custom-agent',
  model: {
    url: 'https://api.siliconflow.cn/v1',
    id: 'siliconflow-cn/ByteDance-Seed/Seed-OSS-36B-Instruct',
    apiKey: process.env.SILICONFLOW_CN_API_KEY,
    headers: {
      'X-Custom-Header': 'value',
    },
  },
})

動態選擇模型
動態選擇模型 的直接連結

src/mastra/agents/my-agent.ts
const agent = new Agent({
  id: 'dynamic-agent',
  name: 'Dynamic Agent',
  model: ({ requestContext }) => {
    const useAdvanced = requestContext.task === 'complex'
    return useAdvanced
      ? 'siliconflow-cn/zai-org/GLM-5.2'
      : 'siliconflow-cn/ByteDance-Seed/Seed-OSS-36B-Instruct'
  },
})
On this page