流式输出

curl -N https://api.tokentosea.com/v1/chat/completions \
  -H "Authorization: Bearer $TOKENTOSEA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{
    "model": "qwen3.8-flash",
    "messages": [{"role": "user", "content": "用 200 字介绍长城。"}],
    "max_tokens": 600,
    "stream": true
  }'

import json
import os
import time
import uuid

import requests

key = str(uuid.uuid4())
started = time.monotonic()
first = last = None
usage = None

with requests.post(
    "https://api.tokentosea.com/v1/chat/completions",
    headers={
        "Authorization": f"Bearer {os.environ['TOKENTOSEA_API_KEY']}",
        "Idempotency-Key": key,
    },
    json={
        "model": "qwen3.8-flash",
        "messages": [{"role": "user", "content": "用 200 字介绍长城。"}],
        "max_tokens": 600,
        "stream": True,
    },
    stream=True,
    timeout=(10, 1300),
) as response:
    if response.status_code != 200:
        raise SystemExit(f"{response.status_code} {response.text}")
    response.encoding = "utf-8"  # text/event-stream 未声明字符集，需显式指定
    event = None
    for line in response.iter_lines(chunk_size=None, decode_unicode=True):
        if not line:
            event = None
            continue
        if line.startswith("event:"):
            event = line[6:].strip()
            continue
        if not line.startswith("data:"):
            continue
        data = line[5:].strip()
        if event == "error":
            raise SystemExit(f"流在结算前中断，请查询 /v1/results/{key}：{data}")
        if data == "[DONE]":
            break
        chunk = json.loads(data)
        if chunk.get("usage"):
            usage = chunk["usage"]
        for choice in chunk.get("choices", []):
            text = (choice.get("delta") or {}).get("content")
            if text:
                now = time.monotonic()
                first = first or now
                last = now
                print(text, end="", flush=True)

print()
if first and usage:
    print(f"首字延迟 {first - started:.2f}s")
    if last > first:
        print(f"TPS {usage['completion_tokens'] / (last - first):.1f}")

import { randomUUID } from 'node:crypto';

const key = randomUUID();
const started = performance.now();
let first, last, usage;

const response = await fetch('https://api.tokentosea.com/v1/chat/completions', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${process.env.TOKENTOSEA_API_KEY}`,
    'Content-Type': 'application/json',
    'Idempotency-Key': key,
  },
  body: JSON.stringify({
    model: 'qwen3.8-flash',
    messages: [{ role: 'user', content: '用 200 字介绍长城。' }],
    max_tokens: 600,
    stream: true,
  }),
});
if (response.status !== 200) throw new Error(`${response.status} ${await response.text()}`);

const decoder = new TextDecoder();
let buffer = '';
let done = false;
for await (const bytes of response.body) {
  buffer = (buffer + decoder.decode(bytes, { stream: true })).replace(/\r\n/g, '\n');
  let end;
  while (!done && (end = buffer.indexOf('\n\n')) !== -1) {
    const frame = buffer.slice(0, end);
    buffer = buffer.slice(end + 2);
    const event = /^event:\s*(.*)$/m.exec(frame)?.[1];
    const data = frame.split('\n').filter((l) => l.startsWith('data:')).map((l) => l.slice(5).trim()).join('\n');
    if (event === 'error') throw new Error(`流在结算前中断，请查询 /v1/results/${key}：${data}`);
    if (!data) continue;
    if (data === '[DONE]') { done = true; break; }
    const chunk = JSON.parse(data);
    if (chunk.usage) usage = chunk.usage;
    for (const choice of chunk.choices ?? []) {
      const text = choice.delta?.content;
      if (!text) continue;
      first ??= performance.now();
      last = performance.now();
      process.stdout.write(text);
    }
  }
  if (done) break;
}
console.log();
if (first && usage) {
  console.log(`首字延迟 ${((first - started) / 1000).toFixed(2)}s`);
  if (last > first) console.log(`TPS ${(usage.completion_tokens / ((last - first) / 1000)).toFixed(1)}`);
}