Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.routing.run/llms.txt

Use this file to discover all available pages before exploring further.

pi.dev is a terminal coding agent with extension support for custom model providers. Use the community pi-routing-run-provider extension to add routing.run, then select any model id that starts with route/.

Setup

1

Install the provider

Install the routing.run extension from GitHub:
pi install git:github.com/monotykamary/pi-routing-run-provider
Confirm it is installed:
pi list
2

Add your API key

Create an API key in app.routing.run. It starts with rk_.Add it to ~/.pi/agent/auth.json:
{
  "routing-run": {
    "type": "api_key",
    "key": "rk_REPLACE_ME"
  }
}
3

Start pi with a routing.run model

Use --provider routing-run and keep the full route/ model id:
pi --provider routing-run --model route/glm-5.1-fp16

One-shot example

Run a single prompt without opening an interactive session:
pi --provider routing-run --model route/kimi-k2.6 -p "Summarize this repository"
For reasoning-capable models, add pi’s thinking flag:
pi --provider routing-run --model route/kimi-k2.6 --thinking high -p "Plan the refactor"

Connection prompt

Copy this prompt when another assistant or teammate is helping you configure pi.dev.

pi.dev - connect to routing.run

Authentication options

The provider can read your key from any of these locations:
MethodUse when
~/.pi/agent/auth.jsonYou want the normal persistent local setup.
--api-keyYou want to pass a key for a single run.
ROUTING_RUN_API_KEYYou prefer environment variables or shell profiles.
Example environment setup:
export ROUTING_RUN_API_KEY='rk_REPLACE_ME'
pi --provider routing-run --model route/glm-5.1-fp16
Published model IDs are listed on the models page. Access depends on your plan tier.
ModelGood for
route/glm-5.1-fp16Strong GLM coding sessions
route/kimi-k2.6Agentic and tool-heavy work
route/glm-5.1-precisionHigher-precision reasoning
route/qwen3.6-plusBalanced coding tasks
route/deepseek-v4-proLong-context reasoning

Troubleshooting

ProblemFix
Provider is missingRun pi install git:github.com/monotykamary/pi-routing-run-provider, then pi list.
Authentication failsCheck that the key starts with rk_ and is stored under routing-run.
Model is missing or staleRun pi update git:github.com/monotykamary/pi-routing-run-provider.
Invalid model errorUse the full model id, for example route/glm-5.1-fp16.