Qwen machine translation - - Alibaba Cloud Documentation Center

Qwen-MT is a large language model for machine translation built based on Qwen. It specializes in Chinese-English translation and multilingual translation between Chinese/English and 24 other languages. Qwen-MT also provides capabilities such as terminology intervention, domain prompting, and translation memory to enhance translation quality in complex scenarios.

Important

If you need to use this model, submit a ticket and make sure to mention "request to access the Qwen-MT model" in the ticket.

Supported models

For higher translation quality, we recommend qwen-mt-plus. For faster speed or lower cost, we recommend qwen-mt-turbo.

Name	Context	Maximum input	Maximum output	Input price	Output price	Free quota (Note)
	(Tokens)			(Million tokens)
qwen-mt-plus	2,048	1,024	1,024	$2.46	$7.37	500,000 tokens each Valid for 180 days after activation
qwen-mt-turbo				$0.16	$0.49

Usage notes

Because of the unique translation scenario, Qwen-MT differs from general text generation models:

You need to pass parameters such as source language (source_lang) and target language (target_lang) through translation_options. If you are not sure about the source language or when the source text contains multiple languages, you can set source_lang to "auto", and the model will automatically detect the source language.
For specific usage, see the code in the following sections.
System Message and multi-round conversation are not supported. The messages array must contain only one User Message, which is the source text.

Currently, DashScope SDK for Java does not support Qwen-MT.

Prerequisites

You must first obtain an API key and set the API key as an environment variable. If you need to use OpenAI SDK or DashScope SDK, you must install the SDK.

Simple example

OpenAI compatible

Here is a simple example of translating from Chinese to English. By referring to Supported languages, you can set source_lang to "Chinese" and target_lang to "English". Then, pass the source text "我看到这个视频后没有笑" in the User Message.

Sample request

Python

import os
from openai import OpenAI

client = OpenAI(
        # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    )
messages = [
      {
        "role": "user",
        "content": "我看到这个视频后没有笑"
      }
    ]
translation_options = {
      "source_lang": "auto",
      "target_lang": "English"
    }
    
completion = client.chat.completions.create(
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    extra_body={
        "translation_options": translation_options
    }
)
print(completion.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "qwen-mt-turbo",
    "messages": [{"role": "user", "content": "看完这个视频我没有笑"}],
    "translation_options": {
      "source_lang": "auto",
      "target_lang": "English"
      }
}'

Sample response

I didn't laugh after watching this video.

DashScope

Sample request

Python

import os
import dashscope

dashscope.base_http_api_url = 'https://dashscope-intl.aliyuncs.com/api/v1'
messages = [
      {
        "role": "user",
        "content": "我看到这个视频后没有笑"
      }
    ]
translation_options = {
      "source_lang": "auto",
      "target_lang": "English",
    }
response = dashscope.Generation.call(
    # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
    api_key=os.getenv('DASHSCOPE_API_KEY'),
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    result_format='message',
    translation_options=translation_options
    )
print(response.output.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/text-generation/generation \
-H "Authorization: $DASHSCOPE_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
  "model": "qwen-mt-turbo",
  "input": {
    "messages": [
      {
        "content": "我看到这个视频后没有笑",
        "role": "user"
      }
    ]
  },
  "parameters": {
    "translation_options": {
      "source_lang": "auto",
      "target_lang": "English"
  }
}'

Sample response

I didn't laugh after watching this video.

Streaming output

In streaming output mode, the model will not generate final response at once. Instead, it gradually generates intermediate results, which need to be joined to form the final response. You can read as the model outputs, thereby shortening the wait for the model's response.

OpenAI compatible

Sample request

Python

import os
from openai import OpenAI

client = OpenAI(
        # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    )
messages = [
      {
        "role": "user",
        "content": "我看到这个视频后没有笑"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English"
    }
    
completion = client.chat.completions.create(
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    stream=True,
    extra_body={
        "translation_options": translation_options
    }
)
for chunk in completion:
    print(chunk.choices[0].delta.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "qwen-mt-turbo",
    "messages": [{"role": "user", "content": "看完这个视频我没有笑"}],
    "stream": true,
    "translation_options": {
      "source_lang": "Chinese",
      "target_lang": "English"
      }
}'

Sample response

I
I didn
I didn't
I didn't laugh after watching this video
I didn't laugh after watching this video.

The Qwen-MT model currently does not support incremental streaming output.

DashScope

Sample request

Python

import os
import dashscope

dashscope.base_http_api_url = 'https://dashscope-intl.aliyuncs.com/api/v1'
messages = [
      {
        "role": "user",
        "content": "我看到这个视频后没有笑"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
    }
response = dashscope.Generation.call(
    # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
    api_key=os.getenv('DASHSCOPE_API_KEY'),
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    result_format='message',
    stream=True,
    translation_options=translation_options
    )
for chunk in response:
    print(chunk.output.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/text-generation/generation \
-H "Authorization: $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-H "X-DashScope-SSE: enable" \
-d '{
  "model": "qwen-mt-turbo",
  "input": {
    "messages": [
      {
        "content": "我看到这个视频后没有笑",
        "role": "user"
      }
    ]
  },
  "parameters": {
    "translation_options": {
      "source_lang": "Chinese",
      "target_lang": "English"
  }
}'

Sample response

I
I didn
I didn't
I didn't laugh after watching this video
I didn't laugh after watching this video.

The Qwen-MT model currently does not support incremental streaming output.

Terminology intervention

If your source text contains many terms, the simple example code may not provide accurate translations. You can translate these terms in advance and provide them to the model as a reference:

Define the term array
Create a JSON array terms that contains your terms. Each term is a JSON object containing the term and its translation:
```
{
    "source": "Term",
    "target": "Pre-translated term"
}
```
Pass the array
Pass the terms array defined in step 1 through translation_options.

After defining and passing the terms, refer to the following code to translate with terminology intervention.

OpenAI compatible

Sample request

Python

import os
from openai import OpenAI

client = OpenAI(
        # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    )
messages = [
      {
        "role": "user",
        "content": "而这套生物传感器运用了石墨烯这种新型材料，它的目标物是化学元素，敏锐的"嗅觉"让它能更深度、准确地体现身体健康状况。"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "terms": [
        {
          "source": "生物传感器",
          "target": "biological sensor"
        },
        {
          "source": "石墨烯",
          "target": "graphene"
        },
        {
          "source": "化学元素",
          "target": "chemical elements"
        },
        {
          "source": "身体健康状况",
          "target": "health status of the body"
        }
      ]
    }
    
completion = client.chat.completions.create(
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    extra_body={
        "translation_options": translation_options
    }
)
print(completion.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
  "model": "qwen-mt-turbo",
  "messages": [
    {
      "role": "user",
      "content": "而这套生物传感器运用了石墨烯这种新型材料，它的目标物是化学元素，敏锐的"嗅觉"让它能更深度、准确地体现身体健康状况。"
    }
  ],
  "translation_options": {
    "source_lang": "Chinese",
    "target_lang": "English",
    "terms": [
      {
        "source": "生物传感器",
        "target": "biological sensor"
      },
      {
        "source": "石墨烯",
        "target": "graphene"
      },
      {
        "source": "化学元素",
        "target": "chemical elements"
      },
      {
        "source": "身体健康状况",
        "target": "health status of the body"
      }
    ]
  }
}'

Sample response

This biological sensor uses graphene, a new material, and its target is chemical elements. Its sensitive "nose" can more deeply and accurately reflect the health status of the body.

DashScope

Sample request

Python

import os
import dashscope

dashscope.base_http_api_url = 'https://dashscope-intl.aliyuncs.com/api/v1'
messages = [
      {
        "role": "user",
        "content": "而这套生物传感器运用了石墨烯这种新型材料，它的目标物是化学元素，敏锐的"嗅觉"让它能更深度、准确地体现身体健康状况。"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "terms": [
        {
          "source": "生物传感器",
          "target": "biological sensor"
        },
        {
          "source": "石墨烯",
          "target": "graphene"
        },
        {
          "source": "化学元素",
          "target": "chemical elements"
        },
        {
          "source": "身体健康状况",
          "target": "health status of the body"
        }
      ]
    }
response = dashscope.Generation.call(
    # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
    api_key=os.getenv('DASHSCOPE_API_KEY'),
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    result_format='message',
    translation_options=translation_options
    )
print(response.output.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/text-generation/generation \
-H "Authorization: $DASHSCOPE_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
  "model": "qwen-mt-turbo",
  "input": {
    "messages": [
      {
        "content": "而这套生物传感器运用了石墨烯这种新型材料，它的目标物是化学元素，敏锐的"嗅觉"让它能更深度、准确地体现身体健康状况。",
        "role": "user"
      }
    ]
  },
  "parameters": {
    "translation_options": {
      "source_lang": "Chinese",
      "target_lang": "English",
      "terms": [
        {
          "source": "生物传感器",
          "target": "biological sensor"
        },
        {
          "source": "石墨烯",
          "target": "graphene"
        },
        {
          "source": "化学元素",
          "target": "chemical elements"
        },
        {
          "source": "身体健康状况",
          "target": "health status of the body"
        }
      ]
  }
}'

Sample response

This biological sensor uses graphene, a new material, and its target is chemical elements. Its sensitive "nose" can more deeply and accurately reflect the health status of the human body.

Translation memory

If you have standard bilingual pairs and need the model to reference these standard translations, you can use the translation memory feature:

Define the translation memory array
Create a JSON array tm_list. Each JSON object contains the source sentence and its standard translation in the following format:
```
{
    "source": "Source sentence",
    "target": "Translated sentence"
}
```
Pass the array
Pass the tm_list array through translation_options.

After defining and passing translation memory, refer to the following code to translate with translation memory.

OpenAI compatible

Sample request

Python

import os
from openai import OpenAI

client = OpenAI(
        # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    )
messages = [
      {
        "role": "user",
        "content": "通过如下命令可以看出安装thrift的版本信息；"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "tm_list":[
          {"source": "您可以通过如下方式查看集群的内核版本信息:", "target": "You can use one of the following methods to query the engine version of a cluster:"},
          {"source": "我们云HBase的thrift环境是0.9.0,所以建议客户端的版本也为 0.9.0,可以从这里下载thrift的0.9.0 版本,下载的源码包我们后面会用到,这里需要先安装thrift编译环境,对于源码安装可以参考thrift官网;", "target": "The version of Thrift used by ApsaraDB for HBase is 0.9.0. Therefore, we recommend that you use Thrift 0.9.0 to create a client. Click here to download Thrift 0.9.0. The downloaded source code package will be used later. You must install the Thrift compiling environment first. For more information, see Thrift official website."},
          {"source": "您可以通过PyPI来安装SDK,安装命令如下:", "target": "You can run the following command in Python Package Index (PyPI) to install Elastic Container Instance SDK for Python:"}
      ]
    }
    
completion = client.chat.completions.create(
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    extra_body={
        "translation_options": translation_options
    }
)
print(completion.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
  "model": "qwen-mt-turbo",
  "messages": [
    {
      "role": "user",
      "content": "通过如下命令可以看出安装thrift的版本信息；"
    }
  ],
  "translation_options": {
    "source_lang": "Chinese",
    "target_lang": "English",
    "tm_list":[
          {"source": "您可以通过如下方式查看集群的内核版本信息:", "target": "You can use one of the following methods to query the engine version of a cluster:"},
          {"source": "我们云HBase的thrift环境是0.9.0,所以建议客户端的版本也为 0.9.0,可以从这里下载thrift的0.9.0 版本,下载的源码包我们后面会用到,这里需要先安装thrift编译环境,对于源码安装可以参考thrift官网;", "target": "The version of Thrift used by ApsaraDB for HBase is 0.9.0. Therefore, we recommend that you use Thrift 0.9.0 to create a client. Click here to download Thrift 0.9.0. The downloaded source code package will be used later. You must install the Thrift compiling environment first. For more information, see Thrift official website."},
          {"source": "您可以通过PyPI来安装SDK,安装命令如下:", "target": "You can run the following command in Python Package Index (PyPI) to install Elastic Container Instance SDK for Python:"}
    ]
  }
}'

Sample response

You can use the following commands to check the version information of thrift installed;

DashScope

Sample request

Python

import os
import dashscope

dashscope.base_http_api_url = 'https://dashscope-intl.aliyuncs.com/api/v1'
messages = [
      {
        "role": "user",
        "content": "通过如下命令可以看出安装thrift的版本信息；"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "tm_list":[
          {"source": "您可以通过如下方式查看集群的内核版本信息:", "target": "You can use one of the following methods to query the engine version of a cluster:"},
          {"source": "我们云HBase的thrift环境是0.9.0,所以建议客户端的版本也为 0.9.0,可以从这里下载thrift的0.9.0 版本,下载的源码包我们后面会用到,这里需要先安装thrift编译环境,对于源码安装可以参考thrift官网;", "target": "The version of Thrift used by ApsaraDB for HBase is 0.9.0. Therefore, we recommend that you use Thrift 0.9.0 to create a client. Click here to download Thrift 0.9.0. The downloaded source code package will be used later. You must install the Thrift compiling environment first. For more information, see Thrift official website."},
          {"source": "您可以通过PyPI来安装SDK,安装命令如下:", "target": "You can run the following command in Python Package Index (PyPI) to install Elastic Container Instance SDK for Python:"}
      ]}
response = dashscope.Generation.call(
    # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
    api_key=os.getenv('DASHSCOPE_API_KEY'),
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    result_format='message',
    translation_options=translation_options
    )
print(response.output.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/text-generation/generation \
-H "Authorization: $DASHSCOPE_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
  "model": "qwen-mt-turbo",
  "input": {
    "messages": [
      {
        "content": "通过如下命令可以看出安装thrift的版本信息；",
        "role": "user"
      }
    ]
  },
  "parameters": {
    "translation_options": {
      "source_lang": "Chinese",
      "target_lang": "English",
      "tm_list":[
          {"source": "您可以通过如下方式查看集群的内核版本信息:", "target": "You can use one of the following methods to query the engine version of a cluster:"},
          {"source": "我们云HBase的thrift环境是0.9.0,所以建议客户端的版本也为 0.9.0,可以从这里下载thrift的0.9.0 版本,下载的源码包我们后面会用到,这里需要先安装thrift编译环境,对于源码安装可以参考thrift官网;", "target": "The version of Thrift used by ApsaraDB for HBase is 0.9.0. Therefore, we recommend that you use Thrift 0.9.0 to create a client. Click here to download Thrift 0.9.0. The downloaded source code package will be used later. You must install the Thrift compiling environment first. For more information, see Thrift official website."},
          {"source": "您可以通过PyPI来安装SDK,安装命令如下:", "target": "You can run the following command in Python Package Index (PyPI) to install Elastic Container Instance SDK for Python:"}
      ]
  }
}'

Sample response

You can use the following commands to check the version information of thrift installed;

Domain prompting

If you want the translation to match the style of a specific domain, such as formal legal language or casual language, provide a natural language description of the domain as a prompt.

Currently, the domain prompt must be in English.

Pass the defined domains prompt through translation_options.

OpenAI compatible

Sample request

Python

import os
from openai import OpenAI

client = OpenAI(
        # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        base_url="https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
    )
messages = [
      {
        "role": "user",
        "content": "第二个SELECT语句返回一个数字，表示在没有LIMIT子句的情况下，第一个SELECT语句返回了多少行。"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "domains": "The sentence is from Alibaba Cloud IT domain. It mainly involves computer-related software development and usage methods, including many terms related to computer software and hardware. Pay attention to professional troubleshooting terminologies and sentence patterns when translating. Translate into this IT domain style."
    }
    
completion = client.chat.completions.create(
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    extra_body={
        "translation_options": translation_options
    }
)
print(completion.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
  "model": "qwen-mt-turbo",
  "messages": [
    {
      "role": "user",
      "content": "第二个SELECT语句返回一个数字，表示在没有LIMIT子句的情况下，第一个SELECT语句返回了多少行。"
    }
  ],
  "translation_options": {
    "source_lang": "Chinese",
    "target_lang": "English",
    "domains": "The sentence is from Alibaba Cloud IT domain. It mainly involves computer-related software development and usage methods, including many terms related to computer software and hardware. Pay attention to professional troubleshooting terminologies and sentence patterns when translating. Translate into this IT domain style."
  }
}'

Sample response

The second SELECT statement returns a number that indicates how many rows were returned by the first SELECT statement without LIMIT clause.

DashScope

Sample request

Python

import os
import dashscope

dashscope.base_http_api_url = 'https://dashscope-intl.aliyuncs.com/api/v1'
messages = [
      {
        "role": "user",
        "content": "第二个SELECT语句返回一个数字，表示在没有LIMIT子句的情况下，第一个SELECT语句返回了多少行。"
      }
    ]
translation_options = {
      "source_lang": "Chinese",
      "target_lang": "English",
      "domains": "The sentence is from Alibaba Cloud IT domain. It mainly involves computer-related software development and usage methods, including many terms related to computer software and hardware. Pay attention to professional troubleshooting terminologies and sentence patterns when translating. Translate into this IT domain style."
      }
response = dashscope.Generation.call(
    # If environment variables are not configured, replace the following line with: api_key="sk-xxx",
    api_key=os.getenv('DASHSCOPE_API_KEY'),
    model="qwen-mt-turbo",  # Using qwen-mt-turbo as an example, you can change the model name as needed
    messages=messages,
    result_format='message',
    translation_options=translation_options
    )
print(response.output.choices[0].message.content)

curl

curl -X POST https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/text-generation/generation \
-H "Authorization: $DASHSCOPE_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
  "model": "qwen-mt-turbo",
  "input": {
    "messages": [
      {
        "content": "第二个SELECT语句返回一个数字，表示在没有LIMIT子句的情况下，第一个SELECT语句返回了多少行。",
        "role": "user"
      }
    ]
  },
  "parameters": {
    "translation_options": {
      "source_lang": "Chinese",
      "target_lang": "English",
      "domains": "The sentence is from Alibaba Cloud IT domain. It mainly involves computer-related software development and usage methods, including many terms related to computer software and hardware. Pay attention to professional troubleshooting terminologies and sentence patterns when translating. Translate into this IT domain style."}
  }
}'

Sample response

The second SELECT statement returns a number that indicates how many rows were returned by the first SELECT statement without a LIMIT clause.

Supported languages

Here are the supported languages and their full names. Use these full names to specify languages in requests.

Language	Full name
Chinese	Chinese
English	English
Japanese	Japanese
Korean	Korean
Thai	Thai
French	French
German	German
Spanish	Spanish
Arabic	Arabic
Indonesian	Indonesian
Vietnamese	Vietnamese
Brazilian Portuguese	Portuguese
Italian	Italian
Dutch	Dutch
Russian	Russian
Khmer	Khmer
Cebuano	Cebuano
Filipino	Filipino
Czech	Czech
Polish	Polish
Persian	Persian
Hebrew	Hebrew
Turkish	Turkish
Hindi	Hindi
Bengali	Bengali
Urdu	Urdu

API reference

For input and output parameters of Qwen-TM, see Qwen.

Error code

If the call failed and an error message is returned, see Error messages.