返回模型列表
Qwen3 32B
qwenqwen/qwen3-32b
API 使用指南
Chat 调用
标准 OpenAI 兼容 Chat Completions。省略 reasoning 参数使用非思考(快速)模式;设置 reasoning: {} 或预算以启用思考模式。
cURL
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-32b",
"messages": [
{"role": "user", "content": "Explain the tradeoffs between dense and MoE architectures for inference."}
],
"temperature": 0.7,
"max_tokens": 2048
}'思考模式
通过 reasoning 参数启用深度链式推理。模型在最终答案前输出 <think>…</think> 块——在展示给终端用户前需解析或过滤。
cURL
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-32b",
"messages": [
{"role": "user", "content": "Prove that the sum of two odd numbers is even."}
],
"reasoning": {},
"temperature": 0.6,
"max_tokens": 8192
}'流式输出
为交互式 UI 流式输出响应。思考模式先流式输出 <think> 块,再输出答案——在读取循环中处理两个片段。
cURL
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-32b",
"messages": [{"role": "user", "content": "Summarize the key points of the Qwen3 technical report."}],
"stream": true
}'工具调用
Qwen3-32B 在思考和非思考两种模式下均支持工具调用——适合需要在工具执行同时保留推理轨迹的 agentic 管道。
cURL
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-32b",
"messages": [{"role": "user", "content": "What is the weather in Shanghai right now?"}],
"tools": [{
"type": "function",
"function": {
"name": "get_weather",
"description": "Get current weather for a location",
"parameters": {
"type": "object",
"properties": {
"location": {"type": "string", "description": "City name"}
},
"required": ["location"]
}
}
}],
"tool_choice": "auto"
}'JSON 模式
JSON 模式适合结构化提取、分类管道和需要机器可读输出的 RAG 排名任务。
cURL
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen3-32b",
"messages": [
{"role": "system", "content": "Return only JSON with keys: summary, sentiment, keywords."},
{"role": "user", "content": "Analyze this product review: The battery life is amazing but the screen is dim."}
],
"response_format": {"type": "json_object"}
}'事实档案 — 本页每条断言可在此回溯来源
| 来源 | URL | 采集于 | |
|---|---|---|---|
| 架构 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 参数规模 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 层数 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 注意力头(GQA) | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 原生上下文窗口 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 思考模式 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 语言支持 | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| 许可证 | huggingface.co ↗ | 2026-06-08 | 已核实 |
| 预训练数据 | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| AIME 2024 | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| MATH-500 | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| LiveCodeBench | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| BFCL (Berkeley Function-Calling Leaderboard) | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| MMLU-Pro | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| C-Eval | qwenlm.github.io ↗ | 2026-06-08 | 已核实 |
| Qwen3 发布 32B 稠密开源权重模型,支持混合思考模式 | Qwen3 announcement ↗ | 2026-06-08 | 已核实 |