pi-dashscope

Alibaba DashScope (ModelStudio) provider for oh my pi — Qwen3, GLM-5, MiniMax M2.5, Kimi K2.5 via OpenAI-compatible coding API

Package details

extension

Install pi-dashscope from npm and Pi will load the resources declared by the package manifest.

$ pi install npm:pi-dashscope

Pi manifest JSON

{
  "extensions": [
    "./index.ts"
  ]
}

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

Alibaba DashScope (ModelStudio) provider for pi.dev — routes Qwen3 / GLM-5 / Kimi K2.5 models through the OpenAI-compatible coding endpoint.

ID	Name	Context	Output	Thinking
`qwen3.5-plus`	Qwen 3.5 Plus	960K	64K	✅
`qwen3-max-2026-01-23`	Qwen3 Max (2026-01-23)	252K	32K	✅
`qwen3-coder-plus`	Qwen3 Coder Plus	975K	64K	❌
`qwen3-coder-next`	Qwen3 Coder Next	200K	64K	❌
`glm-5`	GLM-5	200k	16k	✅
`glm-4.7`	GLM-4.7	166K	16K	✅
`minimax-m2.5`	MiniMax M2.5	192k	32k	✅
`kimi-k2.5`	Kimi K2.5	252K	32K	✅

npm i @mariozechner/pi-coding-agent

# Install using npm package
pi install npm:pi-dashscope

After installation, verify it's working:

# List available models
pi model list

# Check if dashscope models appear:
# dashscope/qwen3.5-plus
# dashscope/qwen3-max-2026-01-23
# etc.

After installing, run the configure command inside pi:

/dashscope-configure

It will prompt you for your API key and save it to ~/.pi/agent/dashscope.json. The key is applied immediately — no restart needed.

Set DASHSCOPE_API_KEY in your environment (takes precedence over the saved config):

export DASHSCOPE_API_KEY=sk-sp-...

Get your API key from the Model Studio console.

After installing, use /model or Ctrl+L in pi to select a dashscope/... model. Use Ctrl+T to toggle thinking for models that support it.