QwenCloud and Qianwen AI Platform (Qwen models)

Alibaba unified its model platform brands (DashScope, Bailian, Model Studio) under QwenCloud outside mainland China and Qianwen AI Platform in mainland China. LiteLLM routes both brands through its existing DashScope implementation, and the legacy dashscope/ prefix keeps working as an alias

Overview

Property	Details
Description	Alibaba's Qwen model platform, served over an OpenAI-compatible API
Provider Routes on LiteLLM	qwencloud/ (international), qwen_ai_platform/ (mainland China), dashscope/ (legacy alias)
Link to Provider Doc	QwenCloud Documentation, Qianwen AI Platform Documentation
Supported Operations	/chat/completions, /embeddings, /rerank, /images/generations

These routes support pay-as-you-go API keys. Token Plan and Coding Plan endpoints are not yet mapped in LiteLLM (tracked on issue #36150)

QwenCloud (international)

Use the qwencloud/ prefix outside mainland China

Setting	Value
Prefix	qwencloud/
Default API base	https://dashscope-intl.aliyuncs.com/compatible-mode/v1
Rerank endpoint	https://dashscope-intl.aliyuncs.com/compatible-api/v1/reranks
Image generation endpoint	https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation
Video generation host	https://dashscope-intl.aliyuncs.com
API key env var	QWENCLOUD_API_KEY (falls back to DASHSCOPE_API_KEY)
Base URL overrides	QWENCLOUD_API_BASE, QWENCLOUD_API_BASE_RERANK, QWENCLOUD_API_BASE_IMAGE, QWENCLOUD_API_BASE_VIDEO

Qianwen AI Platform (mainland China)

Use the qwen_ai_platform/ prefix in mainland China. It hits the same paths on the mainland host

Setting	Value
Prefix	qwen_ai_platform/
Default API base	https://dashscope.aliyuncs.com/compatible-mode/v1
Rerank endpoint	https://dashscope.aliyuncs.com/compatible-api/v1/reranks
Image generation endpoint	https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation
Video generation host	https://dashscope.aliyuncs.com
API key env var	QWEN_AI_PLATFORM_API_KEY (falls back to DASHSCOPE_API_KEY)
Base URL overrides	QWEN_AI_PLATFORM_API_BASE, QWEN_AI_PLATFORM_API_BASE_RERANK, QWEN_AI_PLATFORM_API_BASE_IMAGE, QWEN_AI_PLATFORM_API_BASE_VIDEO

API Key

Environment Variables

import os

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

For mainland China set QWEN_AI_PLATFORM_API_KEY instead. Both prefixes also read DASHSCOPE_API_KEY when their own key is not set, so existing DashScope credentials work without changes

Models

Both prefixes accept any model from the DashScope catalog, for example qwencloud/qwen-max, qwencloud/qwen-flash, qwencloud/qwen-image-3.0, or qwen_ai_platform/qwen-plus. See the DashScope page for the model list

Usage - LiteLLM Python SDK

Chat Completions

QwenCloud Chat Completion

import os
from litellm import completion

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

response = completion(
    model="qwencloud/qwen-max",
    messages=[{"role": "user", "content": "hello from litellm"}],
)

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

Streaming

QwenCloud Streaming Chat Completion

import os
from litellm import completion

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

response = completion(
    model="qwencloud/qwen-flash",
    messages=[{"role": "user", "content": "hello from litellm"}],
    stream=True,
)

for chunk in response:
    print(chunk)

Embedding

QwenCloud Embedding

import os
from litellm import embedding

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

response = embedding(
    model="qwencloud/text-embedding-v4",
    input=["hello from litellm"],
)

print(response.data[0]["embedding"][:5])

Rerank

QwenCloud Rerank

import os
from litellm import rerank

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

response = rerank(
    model="qwencloud/gte-rerank-v2",
    query="What is the capital of France?",
    documents=["Paris is the capital of France", "Berlin is the capital of Germany"],
)

print(response.results)

Image Generation

QwenCloud Image Generation

import os
from litellm import image_generation

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

response = image_generation(
    model="qwencloud/qwen-image-3.0",
    prompt="A cup of coffee on a wooden table",
)

print(response.data[0].url)

Video Generation

Video generation runs as an async task on the DashScope task API. video_generation returns a job id right away; poll video_status until it reports completed, then download the mp4 with video_content.

QwenCloud Video Generation

import os
import time
from litellm import video_content, video_generation, video_status

os.environ["QWENCLOUD_API_KEY"] = "your-api-key"

video = video_generation(
    model="qwencloud/wan3.0-video",
    prompt="A paper boat drifting down a rain-soaked city gutter",
    seconds="5",
    size="1280x720",
)

while video.status not in ("completed", "failed"):
    time.sleep(20)
    video = video_status(video_id=video.id)

with open("video.mp4", "wb") as f:
    f.write(video_content(video_id=video.id))

The supported models are wan3.0-video, wan3.0-video-prime, wan2.7-t2v, wan2.7-i2v, wan2.7-r2v, happyhorse-1.1-t2v, happyhorse-1.1-i2v, happyhorse-1.1-r2v, happyhorse-1.0-t2v, happyhorse-1.0-i2v and happyhorse-1.0-r2v, under any of the qwencloud/, qwen_ai_platform/ and dashscope/ prefixes. Each prefix is priced from the rate card of the region its default host serves. The price follows the prefix, not the host, so a dashscope/ video sent to the international host through DASHSCOPE_API_BASE or DASHSCOPE_API_BASE_VIDEO is still billed at the mainland China rates. Use the qwencloud/ prefix for the international host, or set your own price with custom pricing.

seconds sets the duration in whole seconds. size picks the aspect ratio closest to it out of 16:9, 9:16, 1:1, 4:3 and 3:4, and picks the billed resolution tier from its shorter side: under 600 px is 480P, under 900 px is 720P, and anything larger is 1080P. Wan 2.7 and HappyHorse 1.0 are only offered at 720P and 1080P, so a size in the 480P tier is rejected on them. To get another ratio a model supports, such as 21:9 on Wan 3.0, pass ratio directly. Without size the job runs and bills at 1080P, and without seconds it runs for 5 seconds. litellm only maps size on the video models it has a price for, so on any other DashScope video model a size answers 400 and that model's own size fields go inside parameters instead. A malformed seconds or size, or a resolution the model is not offered at, answers 400 unless drop_params is set globally, on the deployment, or on the request, which drops the value so DashScope's default applies. Image-to-video and reference-to-video models take their image through input_reference, as a URL or as a file (an open file, bytes or a pathlib.Path).

The DashScope fields resolution, ratio, audio, seed, negative_prompt, prompt_extend, watermark and media can go at the top level of the request. Any other DashScope request field, such as shot_type, goes inside a parameters object, which is sent to DashScope as written. From the OpenAI SDK, pass it as extra_body={"parameters": {"shot_type": "multi"}}. A duration, seed, audio, prompt_extend, watermark, resolution or ratio of the wrong type, or a seed outside 0 to 2147483647, answers 400 unless drop_params is set, since DashScope would otherwise fail the task after creating and billing it. Smart duration (duration: -1) and video input (a reference_video entry in media) are rejected, since litellm bills the video when the task is created and their billed length is not known then. Polling a task id DashScope no longer knows (ids expire after 24 hours and only exist in the region that created them) answers 400. DashScope has no list, delete or remix endpoint.

To serve these models on the proxy, set mode: video_generation on the deployment and call /v1/videos:

config.yaml

model_list:
  - model_name: wan3-video
    litellm_params:
      model: qwencloud/wan3.0-video
      api_key: os.environ/QWENCLOUD_API_KEY
    model_info:
      mode: video_generation

Video generation via Proxy - cURL

curl http://localhost:4000/v1/videos \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LITELLM_API_KEY" \
  -d '{
    "model": "wan3-video",
    "prompt": "A paper boat drifting down a rain-soaked city gutter",
    "seconds": "5",
    "size": "1280x720"
  }'

curl http://localhost:4000/v1/videos/$VIDEO_ID \
  -H "Authorization: Bearer $LITELLM_API_KEY"

curl http://localhost:4000/v1/videos/$VIDEO_ID/content \
  -H "Authorization: Bearer $LITELLM_API_KEY" \
  -o video.mp4

Usage - LiteLLM Proxy

Add the models to your LiteLLM Proxy configuration:

config.yaml

model_list:
  - model_name: qwen-max
    litellm_params:
      model: qwencloud/qwen-max
      api_key: os.environ/QWENCLOUD_API_KEY
  - model_name: qwen-plus-cn
    litellm_params:
      model: qwen_ai_platform/qwen-plus
      api_key: os.environ/QWEN_AI_PLATFORM_API_KEY

Start the proxy:

Start LiteLLM Proxy

export QWENCLOUD_API_KEY="your-api-key"
litellm --config config.yaml --port 4000

OpenAI SDK

QwenCloud via Proxy - OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:4000",
    api_key="your-litellm-key",
)

response = client.chat.completions.create(
    model="qwen-max",
    messages=[{"role": "user", "content": "hello from litellm"}],
)

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

cURL

QwenCloud via Proxy - cURL

curl http://localhost:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LITELLM_API_KEY" \
  -d '{
    "model": "qwen-max",
    "messages": [{"role": "user", "content": "hello from litellm"}]
  }'

Backward Compatibility with DashScope

The dashscope/ prefix, DASHSCOPE_API_KEY, DASHSCOPE_API_BASE, and every existing DashScope config keep working unchanged. qwencloud/ and qwen_ai_platform/ are aliases over the same implementation, so no migration is needed: switch prefixes whenever it suits you, or not at all