1

docker pull justsong/one-api:v0.6.10-alpha.1

1
2
3
4
5
6

docker run --name one-api -d \
  --restart always \
  -p 3000:3000 \
  -e TZ=Asia/Shanghai \
  -v /opt/one-api/data:/data \
  justsong/one-api:v0.6.11-alpha.8

1
2
3
4
5
6
7

docker run --name one-api -d \
  --restart always \
  -p 3000:3000 \
  -e SQL_DSN="root:123456@tcp(localhost:3306)/oneapi" \
  -e TZ=Asia/Shanghai \
  -v /opt/one-api/data:/data \
  justsong/one-api:v0.6.11-alpha.8

1
2

docker ps | grep one-api
docker logs one-api

1
2
3
4
5
6
7
8
9
10
11
12
13
14

server{
  server_name oneapi.voidking.com;  # 请根据实际情况修改你的域名

  location / {
    client_max_body_size  64m;
    proxy_http_version 1.1;
    proxy_pass http://localhost:3000;  # 请根据实际情况修改你的端口
    proxy_set_header Host $host;
    proxy_set_header X-Forwarded-For $remote_addr;
    proxy_cache_bypass $http_upgrade;
    proxy_set_header Accept-Encoding gzip;
    proxy_read_timeout 300s;  # GPT-4 需要较长的超时时间，请自行调整
  }
}

1
2
3

{
  "gpt3.5-test": "gpt-3.5-turbo-16k"
}

1

openai.InternalServerError: Error code: 503 - {'error': {'message': '当前分组 default 下对于模型 xxx 无可用渠道 (request id: xxx)', 'type': 'one_api_error'}}

1
2
3

基准价格：0.002 $ /1k tokens
模型倍率 = 输入价格 / 基准价格
补全倍率 = 输出价格 / 输入价格

1
2

模型倍率 = 输入价格 / 基准价格 = (0.18/1000) / 0.002 = 0.09
补全倍率 = 输出价格 / 输入价格 = 0.72 / 0.18 = 4

1

quota = ((prompt_tokens * model_ratio) + (completion_ratio * model_ratio * completion_ratio)) * group_ratio

1

cost = quota * base_price = quota * (0.002/1000)

1
2
3
4
5
6
7
8
9
10
11
12

curl "https://oneapi.voidking.com/v1/chat/completions" \
    -H "Content-Type: application/json" \
    -H "Authorization: Bearer <ONE_API_KEY>" \
    -d '{
        "model": "gpt-3.5-turbo-16k",
        "messages": [
            {
                "role": "user",
                "content": "hello!"
            }
        ]
    }'

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

from openai import OpenAI

client = OpenAI(
  base_url="https://oneapi.voidking.com/v1",
  api_key="<ONE_API_KEY>",
)

completion = client.chat.completions.create(
  model="gpt-3.5-turbo-16k",
  messages=[
    {
      "role": "user",
      "content": "hello!"
    }
  ]
)

print(completion.choices[0].message.content)

1

python test.py