Switch and pick models

kunji writes models as provider:model, for example openai:gpt-4o-mini or groq:llama-3.3-70b-versatile. The provider part is kunji’s id for the company (see Providers); the model part is the provider’s own model name.

See which models a provider has

kunji models openai
kunji models openrouter --search llama

The list comes live from the provider using your key, so it’s always current. Where kunji knows the price, it shows the cost per million input and output tokens.

Set your default model

From the menu: Change default model → pick a provider → type to search → Enter.

Or with a command:

kunji use anthropic:claude-sonnet-4-5

The default is used by kunji chat, kunji ask and saved prompts unless you pass -m.

Switch in the middle of a chat

/model groq:llama-3.3-70b-versatile
✓ Switched to groq:llama-3.3-70b-versatile  conversation kept

The whole conversation so far is sent to the new model, so you can ask one model, then ask another to check or improve the answer. Press Tab after /model to complete model names (models appear after you’ve listed them once with /models or kunji models).

Use a model once

kunji chat -m mistral:mistral-large-latest
kunji ask -m deepseek:deepseek-chat "Explain big-O notation"

Each provider’s default

Every provider you add has its own default model, chosen when you added the key. Writing just the provider uses it:

kunji chat -m groq      # uses your default Groq model