Qwen (Tongyi)

Configure DB-GPT to use Alibaba Cloud's Qwen models via the DashScope API.

Prerequisites

• A DashScope API key
• DB-GPT installed with proxy_openai extra

Install dependencies

uv sync --all-packages \
  --extra "base" \
  --extra "proxy_openai" \
  --extra "rag" \
  --extra "storage_chromadb" \
  --extra "dbgpts"

Configuration

Edit configs/dbgpt-proxy-tongyi.toml:

[models]
[[models.llms]]
name = "qwen-plus"
provider = "proxy/tongyi"
api_base = "https://dashscope.aliyuncs.com/compatible-mode/v1"
api_key = "${env:DASHSCOPE_API_KEY}"

[[models.embeddings]]
name = "text-embedding-v3"
provider = "proxy/tongyi"
api_url = "https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings"
api_key = "${env:DASHSCOPE_API_KEY}"

tip

Set the environment variable to keep your key out of config:

export DASHSCOPE_API_KEY="your-dashscope-api-key"

Available models

LLMs

Model	Config name	Notes
Qwen-Max	qwen-max	Flagship model, best quality
Qwen-Plus	qwen-plus	Balanced performance and cost
Qwen-Turbo	qwen-turbo	Fastest and cheapest
Qwen-Long	qwen-long	Extended context window

Embeddings

Model	Config name	Notes
text-embedding-v3	text-embedding-v3	Recommended
text-embedding-v2	text-embedding-v2	Previous generation

Start the server

uv run dbgpt start webserver --config configs/dbgpt-proxy-tongyi.toml

Troubleshooting

Issue	Solution
InvalidAPIKey	Verify your DashScope API key and ensure it is activated
Model quota exceeded	Check your DashScope console for usage limits
Slow response	Try qwen-turbo for faster responses

What's next

• Getting Started — Full setup walkthrough
• Model Providers — Try other providers