Qwen3 Next 80B

qwenqwen/qwen3-next-80b

Qwen3 Next generation hybrid Transformer-Mamba model (80B total, 3B active MoE with 512 experts). 10x inference throughput vs Qwen3-32B on long contexts.

上下文长度

262K

最大输出

16K

输入价格

$0.240每百万 Tokens

输出价格

$0.960每百万 Tokens

模态能力

文本→文本

价格明细

类型	费率
输入	$0.240 每百万 Tokens
输出	$0.960 每百万 Tokens

支持参数

temperaturemax_tokenstop_ptoolstool_choiceresponse_formatreasoningstop

API 使用示例

cURL

curl https://api.therouter.ai/v1/chat/completions   -H "Content-Type: application/json"   -H "Authorization: Bearer $THE_ROUTER_API_KEY"   -d '{
    "model": "qwen/qwen3-next-80b",
    "messages": [
      {"role": "user", "content": "Summarize the key points from this input."}
    ]
  }'