CLI plugin for Claude Code

One CLI.
Every AI.

Switch between DeepSeek, Gemini, OpenAI, Groq, and more in Claude Code. Install once, works globally in every project.

$npm install -g claude-multimodel

View on npm Browse providers

claude-mm active profile

Active profile

deepseek

deepseek-v4-pro

DeepSeek

$ claude-mm launch deepseek

10+ providers availableactive

10+

providers

config files

command

Up in three commands

No config files. No environment juggling. Works in every project immediately.

Install globally

One install. Slash commands register automatically in Claude Code.

$ npm install -g claude-multimodel

Set your API key

Hidden input keeps keys out of shell history and chat.

$ claude-mm set-key deepseek

Launch with any model

Switch providers in seconds. Relaunch to apply the new profile.

$ claude-mm launch deepseek

Every major provider, built in

Anthropic-compatible providers work directly. OpenAI-only providers bridge via LiteLLM.

NativeCompatibleLiteLLM

anthropicNative

Anthropic

mainclaude-opus-4-7

fastclaude-haiku-4-5

deepseekCompatible

DeepSeek

maindeepseek-v4-pro

fastdeepseek-v4-flash

openaiCompatible

OpenAI

maingpt-5.5

fastgpt-4o-mini

geminiCompatible

Google Gemini

maingemini-2.5-pro

fastgemini-2.5-flash

groqCompatible

Groq

mainllama-3.3-70b-versatile

fastllama-3.1-8b-instant

mistralCompatible

Mistral AI

mainmistral-large-latest

fastmistral-small-latest

togetherCompatible

Together AI

mainllama-3.3-70b-turbo

fastllama-3.2-3b-turbo

qwenCompatible

Qwen (Alibaba)

mainqwen-max

fastqwen-turbo

openrouterLiteLLM

OpenRouter

main100+ models

fastany

ollamaCompatible

Ollama

mainllama3.3:70b

fastllama3.2:3b

kimiLiteLLM

Kimi (Moonshot)

mainmoonshot-v1-128k

fastmoonshot-v1-8k

glmLiteLLM

GLM (Zhipu AI)

mainglm-4-plus

fastglm-4-flash

LiteLLM providers require a local proxy running on port 4000. Run: litellm --model openrouter/... --port 4000

Every workflow, one command away

Launch Claude with any provider, make it permanent, update model IDs, or browse the full provider registry.

claude-mm launch [profile]

claude-mm set-default [profile]

claude-mm set-model [profile] [model]

claude-mm set-key [profile]

claude-mm add

$claude-mm launch deepseek

Profile: deepseek — DeepSeek v4 Pro

Model: deepseek-v4-pro Effort: max

Starting claude...

Built for how developers actually work

From quick session switches to permanent global defaults. Every workflow covered.

Make any provider permanent

Write env vars directly to your shell profile. Even plain claude uses your chosen provider without touching claude-mm.

$ claude-mm set-default deepseek

> Written to $PROFILE

> Restart terminal. DeepSeek is now your default.

Keys stored safely

Hidden terminal input. Never pasted in chat, never in shell history. Stored locally in ~/.claude/multimodel/.

$ claude-mm set-key deepseek

Slash commands inside Claude

Use /multimodel/switch-profile from within any Claude Code session.

/multimodel/switch-profile

/multimodel/set-api-key

/multimodel/add-profile

Works in every project

Install once globally. Profiles and slash commands available in all Claude Code sessions across every directory.

Add any model or endpoint

Custom base URLs, local Ollama instances, or any Anthropic-compatible endpoint.

$ claude-mm add

Start with one command

Free. Open source. Works with Claude Code on Windows, Mac, and Linux.