Overview
MMX-CLI is the official command-line interface for the MiniMax AI platform. It exposes MiniMax's generative models as terminal commands — multi-turn chat with streaming and JSON output, text-to-image with aspect-ratio and batch controls, asynchronous video generation with progress tracking, text-to-speech across 30+ voices, image understanding, and MiniMax-powered web search.
The tool is built for AI agents as much as for humans. It installs either as a global npm binary for terminal use or as an agent skill, so a coding agent such as OpenClaw, Cursor or Claude Code can call MiniMax's media models directly as shell commands rather than through an MCP integration. Every command is designed to be scriptable: prompts can come from flags, from stdin, or from a messages file, and output can be forced to JSON.
Region handling is built in. MMX-CLI speaks to both the global platform at `api.minimax.io` and the CN platform at `api.minimaxi.com`, using a saved or automatically detected region rather than requiring you to pick per call. Video generation covers both the Hailuo-2.3 and MiniMax-H3 model paths, with local files and Base64 data URIs preflight-checked against the documented size limits before the request goes out. It requires Node.js 18 or later and a MiniMax Token Plan.
What it does
- Seven capabilities from one binary: text chat, image, video, speech, vision, search and quota
- Installs as a global npm CLI or as an agent skill for OpenClaw, Cursor and Claude Code
- Multi-turn chat with streaming, system prompts and JSON output; messages can be piped in from a file or stdin
- Asynchronous video generation with task polling and download, across Hailuo-2.3 and MiniMax-H3
- Text-to-speech with 30+ voices, speed control and streaming playback
- Preflight checks on local media and Base64 data URIs against documented size limits before upload
- Dual-region support for the global and CN MiniMax platforms with automatic region detection
Getting started
MMX-CLI needs Node.js 18+ and a MiniMax Token Plan on either the global or CN platform. Install it as an agent skill if you want an AI coding agent to use it, or globally if you want it in your own shell.
Install
Add it as a skill for an AI agent, or install the CLI globally for terminal use.
# For AI agents (OpenClaw, Cursor, Claude Code, etc.)
npx skills add MiniMax-AI/cli -y -g
# Or install the CLI globally
npm install -g mmx-cliAuthenticate
Log in interactively and choose MiniMax OAuth or paste an API key, or pass the key directly for non-interactive setups.
mmx auth login
# or
mmx auth login --api-key sk-xxxxxGenerate something
Each modality is its own subcommand. These are the README's own quick-start examples.
mmx text chat --message "What is MiniMax?"
mmx image "A cat in a spacesuit"
mmx speech synthesize --text "Hello!" --out hello.mp3
mmx video generate --prompt "Ocean waves at sunset"
mmx search "MiniMax AI latest news"
mmx vision photo.jpg
mmx quotaScript it
Chat accepts role-prefixed messages, a messages file (including stdin), a model override and JSON output — enough to drive it from a pipeline rather than by hand.
mmx text chat --model MiniMax-M3 --message "Hello" --stream
mmx text chat --system "You are a coding assistant" --message "Fizzbuzz in Go"
cat messages.json | mmx text chat --messages-file - --output jsonWork with async video
Video generation returns a task you poll and then download. MiniMax-H3 supports start images, reference images and reference video.
mmx video generate --prompt "A robot painting" --async
mmx video task get --task-id 123456
mmx video download --file-id 176844028768320 --out video.mp4Commands and code are distilled from the project's own documentation — always check the official repo for the latest.
When to use it
- Let a coding agent generate images, video or narration as a shell command, without wiring up an MCP server
- Script bulk media generation — batches of images or narrated clips — from a CI job or a Makefile
- Prototype against MiniMax's text, vision and search models from the terminal before committing to an SDK
- Run the same automation against both the global and CN MiniMax platforms without changing the commands
How MiniMax MMX-CLI compares
MiniMax MMX-CLI alongside other open-source media model apis tools AI/TLDR tracks, ranked by GitHub stars.
| Tool | Stars | What it does |
|---|---|---|
| MiniMax MMX-CLI | ★ 2.1k | Official MiniMax CLI that gives agents and terminals text, image, video, speech and search |
| ElevenLabs | — | AI voice platform with APIs for text-to-speech, voice cloning, dubbing, speech-to-text, and music generation across 70+ languages. |
| Deepgram | — | Enterprise voice AI APIs for real-time speech-to-text, text-to-speech, and voice agent orchestration. |
| AssemblyAI | — | Speech-to-text and speech understanding API that transcribes audio and extracts insights like sentiment and entities. |
| Stability AI | — | Generative media API for image, video, audio, and 3D generation, including Stable Diffusion and Stable Audio models. |
| Luma AI | — | Dream Machine API for generating video and images from text and image prompts. |
| fal.ai | — | Generative media platform serving 1,000+ image, video, audio, and 3D models through a fast unified inference API. |
| Hume AI | — | Expressive voice AI APIs including the Octave TTS model and Empathic Voice Interface for emotionally aware speech. |