Skip to content

README.md

Latest commit

 

History

History
182 lines (129 loc) · 4.2 KB

README.md

File metadata and controls

182 lines (129 loc) · 4.2 KB
Raw

minimax-cli

CLI for the MiniMax AI Platform — generate text, images, video, speech, and music from the terminal.

Supports Global (api.minimax.io) and CN (api.minimaxi.com) with automatic region detection.

中文文档

Install

Binary (recommended) — macOS / Linux / Windows:

curl -fsSL https://raw.githubusercontent.com/MiniMax-AI-Dev/cli/main/install.sh | sh

npm / bun:

npm install -g minimax-cli

Quick Start

minimax auth login --api-key sk-xxxxx

minimax text chat --message "What is MiniMax?"
minimax image "A cat in a spacesuit"
minimax speech synthesize --text "Hello!" --out hello.mp3
minimax search "MiniMax AI latest news"
minimax vision photo.jpg
minimax quota

Commands

Text

minimax text chat --message "Write a poem"
minimax text chat --model MiniMax-M2.7-highspeed --message "Hello" --stream
minimax text chat --system "You are a coding assistant" --message "Fizzbuzz in Go"
minimax text chat --message "user:Hi" --message "assistant:Hey!" --message "How are you?"
cat messages.json | minimax text chat --messages-file - --output json

Image

minimax image "A cat in a spacesuit"              # shorthand
minimax image generate --prompt "A cat" --n 3 --aspect-ratio 16:9
minimax image generate --prompt "Logo" --out-dir ./out/

Video

minimax video generate --prompt "Ocean waves at sunset" --async
minimax video generate --prompt "A robot painting" --download sunset.mp4
minimax video task get --task-id 123456
minimax video download --file-id 176844028768320 --out video.mp4

Speech

minimax speech synthesize --text "Hello!" --out hello.mp3
minimax speech synthesize --text "Stream me" --stream | mpv -
minimax speech synthesize --text "Hi" --voice Boyan_new_hailuo --speed 1.2
echo "Breaking news" | minimax speech synthesize --text-file - --out news.mp3
minimax speech voices

Music

minimax music "Upbeat pop"                        # shorthand
minimax music generate --prompt "Jazz" --lyrics "La la la" --out song.mp3

Vision

minimax vision photo.jpg                          # shorthand
minimax vision describe --image https://example.com/img.jpg --prompt "What breed?"
minimax vision describe --file-id file-123

Search

minimax search "MiniMax AI"                       # shorthand
minimax search query --q "latest news" --output json

Quota & Config

minimax quota
minimax config show
minimax config set --key region --value cn
minimax config export-schema | jq .

Auth

minimax auth login --api-key sk-xxxxx
minimax auth login                    # OAuth browser flow
minimax auth status
minimax auth refresh
minimax auth logout

Update

minimax update
minimax update latest

Global Flags

Flag Description
--api-key <key> API key (overrides config)
--region <global|cn> API region
--output <text|json|yaml> Output format
--quiet Data only on stdout, no decorations
--verbose Print HTTP request/response
--dry-run Show request body, no API call
--async Return task ID immediately (video)
--non-interactive Disable prompts (CI/agent mode)
--timeout <seconds> Request timeout (default: 300)
--no-color Disable ANSI colors

Add --help to any command for full options.

Agent / CI Integration

Export all commands as JSON Tool Schema in one shot:

minimax config export-schema | jq .
minimax config export-schema --command "text chat"

Compatible with Cursor, Cline, Dify, and any framework that speaks Anthropic/OpenAI tool schemas. See skill/SKILL.md for the full integration guide.

Piping:

minimax text chat --message "Hello" --output json | jq .content
minimax image "A logo" --quiet | xargs curl -O
minimax video generate --prompt "A robot" --async --quiet

stdout is always clean data. stderr carries all UI (status bar, spinners, progress).

Output Philosophy

  • stdout → data only (text, URLs, JSON, raw audio bytes)
  • stderr → status bar, progress, warnings, help

License

MIT