流式输出
传入 stream: true 即可让响应以 Server-Sent Events 流的形式逐 token 推送,适合实时打字机效果。
请求
bash
curl https://onegate.club/v1/chat/completions \
-H "Authorization: Bearer $ONEGATE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-4-6",
"messages": [{"role": "user", "content": "Write a haiku."}],
"stream": true
}'事件流
每个事件以 data: 开头,承载一个增量 delta。流末尾发送 data: [DONE]。
text
data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{"role":"assistant","content":""}}]}
data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{"content":"Soft"}}]}
data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{"content":" cherry"}}]}
data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{"content":" blossoms..."}}]}
data: [DONE]Python 客户端
python
stream = client.chat.completions.create(
model="claude-sonnet-4-6",
messages=[{"role": "user", "content": "Tell a short story."}],
stream=True,
)
for chunk in stream:
delta = chunk.choices[0].delta.content or ""
print(delta, end="", flush=True)Node.js 客户端
javascript
const stream = await client.chat.completions.create({
model: "gpt-5.1",
messages: [{ role: "user", content: "Tell a short story." }],
stream: true,
});
for await (const chunk of stream) {
process.stdout.write(chunk.choices[0].delta.content ?? "");
}取消与计费结算
客户端取消下游流后,OneGate 会继续读取上游;如果取得最终实际 usage,仍会按该 usage 正常收费。
cancel-only drain 最长 600 秒;超时后会取消上游、零扣费并记录失败。正常活动中的流不受这项仅取消后启动的计时器影响。
若流式计费写入失败,下游流会终止并记录失败,系统不会自动重试计费。
流式响应也计入 token 用量。stream_options.include_usage = true 可让最后一条事件包含完整 usage 字段。