Chat Completions | Tutujin API Docs

POST /v1/chat/completions is the endpoint you will use most: send a list of messages, get the model's reply. The format matches OpenAI Chat Completions, and every model the catalog lists for this endpoint — GPT, Claude, Gemini, DeepSeek, Qwen and others — works with the same code. We have verified this endpoint on this site, streaming included.

POSThttps://open.tutujin.com/v1/chat/completions

Headers

Header	Required	Description
`Authorization`	Yes	`Bearer sk-...`, an API key from the console
`Content-Type`	Yes	`application/json`

Request body

Parameter	Type	Required	Description
`model`	string	Yes	Model name, e.g. `gpt-5.4-mini`
`messages`	array	Yes	The conversation so far, oldest first — structure below
`stream`	boolean	No	`true` streams the reply as server-sent events; default `false`
`stream_options`	object	No	With streaming, `{"include_usage": true}` adds usage to the last chunk
`max_tokens`	integer	No	Most tokens to generate. Some newer models use `max_completion_tokens` instead
`max_completion_tokens`	integer	No	Generation limit, including reasoning tokens on reasoning models
`temperature`	number	No	Sampling temperature, usually 0 to 2; higher is more varied. Reasoning models usually ignore or reject it
`top_p`	number	No	Nucleus sampling; tune this or `temperature`, not both
`n`	integer	No	Number of candidate replies; each one is billed for the tokens it generates
`stop`	string or array	No	Stop generating when one of these strings appears
`presence_penalty`	number	No	-2 to 2; positive values push the model toward new topics
`frequency_penalty`	number	No	-2 to 2; positive values reduce repeated wording
`response_format`	object	No	`{"type": "json_object"}` asks for JSON; some models support `json_schema` for strict structured output
`tools`	array	No	Function definitions the model may call (function calling)
`tool_choice`	string or object	No	`auto`, `none`, `required`, or a specific function
`reasoning_effort`	string	No	Reasoning depth such as `low`, `medium`, `high`; only some models support it
`seed`	integer	No	Best-effort determinism for identical inputs; not guaranteed
`user`	string	No	Your end-user identifier, useful when investigating abuse

Vendors support optional parameters to different degrees. A parameter a model does not support may be ignored or may return a 400, so try new models with a little traffic first.

The messages array

Each message is an object whose role is one of:

role	Use
`system`	System prompt: role, rules and output format; usually the first message
`user`	User input
`assistant`	The model's earlier replies, sent back as-is in multi-turn chats
`tool`	The result of a function call, with the matching `tool_call_id`

content is a string or an array. Use the array form for mixed text and images on models that support vision:

JSON

{
  "role": "user",
  "content": [
    { "type": "text", "text": "What is the total on this invoice?" },
    { "type": "image_url", "image_url": { "url": "https://example.com/invoice.png" } }
  ]
}

You can also inline an image in url as data:image/png;base64,....

Request example

curl https://open.tutujin.com/v1/chat/completions \
  -H "Authorization: Bearer $TUTUJIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4-mini",
    "messages": [
      {"role": "system", "content": "You are a store support agent. Keep answers short and polite."},
      {"role": "user", "content": "How soon do orders ship?"}
    ],
    "temperature": 0.3
  }'

import os
from openai import OpenAI

client = OpenAI(base_url="https://open.tutujin.com/v1", api_key=os.environ["TUTUJIN_API_KEY"])

resp = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[
        {"role": "system", "content": "You are a store support agent. Keep answers short and polite."},
        {"role": "user", "content": "How soon do orders ship?"},
    ],
    temperature=0.3,
)
print(resp.choices[0].message.content)
print(resp.usage)

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://open.tutujin.com/v1', apiKey: process.env.TUTUJIN_API_KEY })

const resp = await client.chat.completions.create({
  model: 'gpt-5.4-mini',
  messages: [
    { role: 'system', content: 'You are a store support agent. Keep answers short and polite.' },
    { role: 'user', content: 'How soon do orders ship?' },
  ],
  temperature: 0.3,
})
console.log(resp.choices[0].message.content)
console.log(resp.usage)

Response example

JSON

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1794000000,
  "model": "gpt-5.4-mini",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "In-stock orders usually ship within 48 hours of payment."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 31,
    "completion_tokens": 22,
    "total_tokens": 53
  }
}

Field	Description
`choices[].message.content`	The model's answer
`choices[].message.tool_calls`	Present when the model decides to call a function: function name and JSON arguments
`choices[].finish_reason`	`stop` finished normally; `length` hit the generation limit; `tool_calls` wants you to run a function; `content_filter` was blocked by a safety policy
`usage.prompt_tokens`	Input tokens, billed at the input price
`usage.completion_tokens`	Output tokens (including reasoning on reasoning models), billed at the output price
`usage.prompt_tokens_details.cached_tokens`	Input tokens served from the context cache, when the model supports caching; billed at the cache-read price

Streaming

With "stream": true the server sends server-sent events. Each chunk is one data: {...} line with new text in choices[0].delta.content, and the stream ends with data: [DONE]:

Text

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant","content":""}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"In-stock"}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

With an SDK, iterate over the stream:

stream = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[{"role": "user", "content": "Write a 60-word product description"}],
    stream=True,
    stream_options={"include_usage": True},
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)
    if chunk.usage:
        print("\n", chunk.usage)

const stream = await client.chat.completions.create({
  model: 'gpt-5.4-mini',
  messages: [{ role: 'user', content: 'Write a 60-word product description' }],
  stream: true,
  stream_options: { include_usage: true },
})
for await (const chunk of stream) {
  const text = chunk.choices[0]?.delta?.content
  if (text) process.stdout.write(text)
  if (chunk.usage) console.log('\n', chunk.usage)
}

Turn streaming on for long outputs: users see text sooner, and a single long wait is less likely to be cut off by a gateway or client timeout.

Notes

Model names must match the model hub exactly; they are case-sensitive.
A key only uses the lines of its own group. If a model returns 503 saying no channel is available, that group usually does not offer the model — try a key in another group. See Choosing a group.
Reasoning tokens count as output tokens, so reasoning models can cost noticeably more. Cap them with max_completion_tokens.
The server keeps no conversation state. Send the history with every request; longer history means more input tokens, so trim it when you can.
Tokens and cost for every request are listed under Usage in the console.

Models that support this endpoint

Model	Vendor	Cheapest group	Input ¥/1M	Output ¥/1M	Per request ¥
gpt-image-2	OpenAI	Standard	—	—	¥0.12
Claude Haiku 5.5	Anthropic	Claude Plus	¥0.4	¥2	—
qwen3.5-plus	Qwen	China Models	¥1.12	¥6.72	—
gemini-3.8-flash	Google	Standard	¥1.5	¥7.5	—
gpt-5.4-mini	OpenAI	Standard	¥1.5	¥9	—
deepseek-v4.1-flash	DeepSeek	China Models	¥2.8	¥11.2	—
Claude Sonnet 5.5	Anthropic	Claude Lite	¥3.2	¥16	—
GPT-6.1 Sol	OpenAI	Standard	¥4	¥20	—
gemini-3.1-pro-preview	Google	Standard	¥4	¥24	—
Claude Opus 5.5	Anthropic	Claude Lite	¥8	¥40	—

Prices in CNY, cheapest group for each model; updated Nov 12, 2026, 8:17 PM.

681 models support this endpoint; the table shows some of them. See them all in the model hub

Still stuck? Add our WeChat support JsnonoChat and send the request id from the error message, and we will look into it.

POSThttps://open.tutujin.com/v1/chat/completions

Headers

Header	Required	Description
`Authorization`	Yes	`Bearer sk-...`, an API key from the console
`Content-Type`	Yes	`application/json`

Request body

Parameter	Type	Required	Description
`model`	string	Yes	Model name, e.g. `gpt-5.4-mini`
`messages`	array	Yes	The conversation so far, oldest first — structure below
`stream`	boolean	No	`true` streams the reply as server-sent events; default `false`
`stream_options`	object	No	With streaming, `{"include_usage": true}` adds usage to the last chunk
`max_tokens`	integer	No	Most tokens to generate. Some newer models use `max_completion_tokens` instead
`max_completion_tokens`	integer	No	Generation limit, including reasoning tokens on reasoning models
`temperature`	number	No	Sampling temperature, usually 0 to 2; higher is more varied. Reasoning models usually ignore or reject it
`top_p`	number	No	Nucleus sampling; tune this or `temperature`, not both
`n`	integer	No	Number of candidate replies; each one is billed for the tokens it generates
`stop`	string or array	No	Stop generating when one of these strings appears
`presence_penalty`	number	No	-2 to 2; positive values push the model toward new topics
`frequency_penalty`	number	No	-2 to 2; positive values reduce repeated wording
`response_format`	object	No	`{"type": "json_object"}` asks for JSON; some models support `json_schema` for strict structured output
`tools`	array	No	Function definitions the model may call (function calling)
`tool_choice`	string or object	No	`auto`, `none`, `required`, or a specific function
`reasoning_effort`	string	No	Reasoning depth such as `low`, `medium`, `high`; only some models support it
`seed`	integer	No	Best-effort determinism for identical inputs; not guaranteed
`user`	string	No	Your end-user identifier, useful when investigating abuse

Vendors support optional parameters to different degrees. A parameter a model does not support may be ignored or may return a 400, so try new models with a little traffic first.

The messages array

Each message is an object whose role is one of:

role	Use
`system`	System prompt: role, rules and output format; usually the first message
`user`	User input
`assistant`	The model's earlier replies, sent back as-is in multi-turn chats
`tool`	The result of a function call, with the matching `tool_call_id`

content is a string or an array. Use the array form for mixed text and images on models that support vision:

JSON

{
  "role": "user",
  "content": [
    { "type": "text", "text": "What is the total on this invoice?" },
    { "type": "image_url", "image_url": { "url": "https://example.com/invoice.png" } }
  ]
}

You can also inline an image in url as data:image/png;base64,....

Request example

curl https://open.tutujin.com/v1/chat/completions \
  -H "Authorization: Bearer $TUTUJIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4-mini",
    "messages": [
      {"role": "system", "content": "You are a store support agent. Keep answers short and polite."},
      {"role": "user", "content": "How soon do orders ship?"}
    ],
    "temperature": 0.3
  }'

import os
from openai import OpenAI

client = OpenAI(base_url="https://open.tutujin.com/v1", api_key=os.environ["TUTUJIN_API_KEY"])

resp = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[
        {"role": "system", "content": "You are a store support agent. Keep answers short and polite."},
        {"role": "user", "content": "How soon do orders ship?"},
    ],
    temperature=0.3,
)
print(resp.choices[0].message.content)
print(resp.usage)

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://open.tutujin.com/v1', apiKey: process.env.TUTUJIN_API_KEY })

const resp = await client.chat.completions.create({
  model: 'gpt-5.4-mini',
  messages: [
    { role: 'system', content: 'You are a store support agent. Keep answers short and polite.' },
    { role: 'user', content: 'How soon do orders ship?' },
  ],
  temperature: 0.3,
})
console.log(resp.choices[0].message.content)
console.log(resp.usage)

Response example

JSON

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1794000000,
  "model": "gpt-5.4-mini",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "In-stock orders usually ship within 48 hours of payment."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 31,
    "completion_tokens": 22,
    "total_tokens": 53
  }
}

Field	Description
`choices[].message.content`	The model's answer
`choices[].message.tool_calls`	Present when the model decides to call a function: function name and JSON arguments
`choices[].finish_reason`	`stop` finished normally; `length` hit the generation limit; `tool_calls` wants you to run a function; `content_filter` was blocked by a safety policy
`usage.prompt_tokens`	Input tokens, billed at the input price
`usage.completion_tokens`	Output tokens (including reasoning on reasoning models), billed at the output price
`usage.prompt_tokens_details.cached_tokens`	Input tokens served from the context cache, when the model supports caching; billed at the cache-read price

Streaming

With "stream": true the server sends server-sent events. Each chunk is one data: {...} line with new text in choices[0].delta.content, and the stream ends with data: [DONE]:

Text

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"role":"assistant","content":""}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{"content":"In-stock"}}]}

data: {"id":"chatcmpl-...","object":"chat.completion.chunk","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

With an SDK, iterate over the stream:

stream = client.chat.completions.create(
    model="gpt-5.4-mini",
    messages=[{"role": "user", "content": "Write a 60-word product description"}],
    stream=True,
    stream_options={"include_usage": True},
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)
    if chunk.usage:
        print("\n", chunk.usage)

const stream = await client.chat.completions.create({
  model: 'gpt-5.4-mini',
  messages: [{ role: 'user', content: 'Write a 60-word product description' }],
  stream: true,
  stream_options: { include_usage: true },
})
for await (const chunk of stream) {
  const text = chunk.choices[0]?.delta?.content
  if (text) process.stdout.write(text)
  if (chunk.usage) console.log('\n', chunk.usage)
}

Turn streaming on for long outputs: users see text sooner, and a single long wait is less likely to be cut off by a gateway or client timeout.

Notes

Model names must match the model hub exactly; they are case-sensitive.
A key only uses the lines of its own group. If a model returns 503 saying no channel is available, that group usually does not offer the model — try a key in another group. See Choosing a group.
Reasoning tokens count as output tokens, so reasoning models can cost noticeably more. Cap them with max_completion_tokens.
The server keeps no conversation state. Send the history with every request; longer history means more input tokens, so trim it when you can.
Tokens and cost for every request are listed under Usage in the console.

Models that support this endpoint

Model	Vendor	Cheapest group	Input ¥/1M	Output ¥/1M	Per request ¥
gpt-image-2	OpenAI	Standard	—	—	¥0.12
Claude Haiku 5.5	Anthropic	Claude Plus	¥0.4	¥2	—
qwen3.5-plus	Qwen	China Models	¥1.12	¥6.72	—
gemini-3.8-flash	Google	Standard	¥1.5	¥7.5	—
gpt-5.4-mini	OpenAI	Standard	¥1.5	¥9	—
deepseek-v4.1-flash	DeepSeek	China Models	¥2.8	¥11.2	—
Claude Sonnet 5.5	Anthropic	Claude Lite	¥3.2	¥16	—
GPT-6.1 Sol	OpenAI	Standard	¥4	¥20	—
gemini-3.1-pro-preview	Google	Standard	¥4	¥24	—
Claude Opus 5.5	Anthropic	Claude Lite	¥8	¥40	—

Prices in CNY, cheapest group for each model; updated Nov 12, 2026, 8:17 PM.

681 models support this endpoint; the table shows some of them. See them all in the model hub