Publish time
-Model Series
千问Input type
Output type
Input Price
¥2 / 1M tokensOutput Price
¥20 / 1M tokensCache Write Price
¥2.5 / 1M tokensCache Read Price
¥0.2 / 1M tokensContext Window
128,000Max Output Length
31,000实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。
Zhinao API routes requests to the best-fit provider and automatically fails over to the one with highest availability.
TTFT
No data
Throughput
44.18tps
Uptime
100.00%
Provider Model
huaweicloud/qwen/qwen3-32b
Supported Parameters
Recent Uptime
Reasoning
Toggleable
Supported Response Formats
Total Context
128,000
Max Output
8,000
Input Price
¥1.06 / 1M tokens
Output Price
¥10.6 / 1M tokens
Cache Write
¥1.325 / 1M tokens
Cache Read
¥0.106 / 1M tokens
TTFT
0.02s
Throughput
70.28tps
Uptime
100.00%
Provider Model
qiniu/qwen3-32b
Supported Parameters
Recent Uptime
Reasoning
-
Supported Response Formats
Total Context
65,536
Max Output
31,000
Input Price
¥2 / 1M tokens
Output Price
¥20 / 1M tokens
Cache Write
¥2.5 / 1M tokens
Cache Read
¥0.2 / 1M tokens
TTFT
No data
Throughput
37tps
Uptime
100.00%
Provider Model
alibaba/qwen3-32b
Supported Parameters
Recent Uptime
Reasoning
-
Supported Response Formats
Total Context
65,536
Max Output
31,000
Input Price
¥2 / 1M tokens
Output Price
¥20 / 1M tokens
Cache Write
¥2.5 / 1M tokens
Cache Read
¥0.2 / 1M tokens
Compare different providers across Zhinao API
65.47 tok/s
0.03 s
Uptime for qwen3-32b across all providers
Zhinao API normalizes requests and responses across providers for you
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.360.cn/v1",
apiKey: process.env.ZHINAO_API_KEY,
});
const response = await client.chat.completions.create({
model: "qwen3-32b",
messages: [
{ role: "user", content: "Hello, how are you?" }
],
temperature: 0.7,
max_tokens: 1000,
});
console.log(response.choices[0].message.content);