Kimi CLI

Check Kimi CLI

Run:

kimi --version

If it is not installed yet, install Kimi CLI using the official installation flow before continuing.

Set API Key environment variables

Because the provider type below is openai_legacy, Kimi CLI reads OPENAI_API_KEY and OPENAI_BASE_URL for environment overrides.

export OPENAI_API_KEY="sk-yolo-..."
export OPENAI_BASE_URL="https://api.yolorouter.com/v1"

$env:OPENAI_API_KEY = "sk-yolo-..."
$env:OPENAI_BASE_URL = "https://api.yolorouter.com/v1"

set OPENAI_API_KEY=sk-yolo-...
set OPENAI_BASE_URL=https://api.yolorouter.com/v1

Persist them in your shell profile or system environment for regular use. Do not commit the real API Key to your repository.

Edit the config file

Kimi CLI's default config file is:

~/.kimi/config.toml

On Windows this is usually:

%USERPROFILE%\.kimi\config.toml

You can also type /config inside Kimi CLI to open the config file quickly.

Add the Yolo Router provider

Add or merge the following config:

default_model = "deepseek-v4-pro"

[providers.yolorouter]
type = "openai_legacy"
base_url = "https://api.yolorouter.com/v1"
api_key = ""

[models.deepseek-v4-pro]
provider = "yolorouter"
model = "deepseek-v4-pro"
max_context_size = 128000
capabilities = ["thinking"]
display_name = "DeepSeek V4 Pro"

[models.deepseek-v4-flash]
provider = "yolorouter"
model = "deepseek-v4-flash"
max_context_size = 128000
display_name = "DeepSeek V4 Flash"

Field notes:

• default_model: the model Kimi CLI uses by default; it must be a key defined under [models.*]
• type: use openai_legacy for OpenAI Chat Completions compatible services
• base_url: your Yolo Router OpenAI-compatible endpoint, with /v1
• api_key: left empty in this example because the real Key is injected through OPENAI_API_KEY
• models.*.provider: must point to the yolorouter provider above
• models.*.model: must exactly match the model ID in the Yolo Router console
• capabilities = ["thinking"]: only set this for models that support reasoning/thinking mode; remove it for normal models

Start and select a model

Open your project directory and run:

kimi

Inside Kimi CLI, type:

/model

Select deepseek-v4-pro or another Yolo Router model you configured.

You can also select the model at startup:

kimi --model deepseek-v4-pro

Test the connection

In interactive mode, send:

Introduce the model you are using in one sentence.

Or run a one-off test with Print mode:

kimi --print "Introduce the model you are using in one sentence"

If Kimi CLI returns a normal response, the integration is working. You can also check the Yolo Router console for the corresponding request log.