RunAPI CLI.
AI Models in Your Terminal.
One binary calls 130+ AI models — Suno, Veo, Kling, Flux, Claude, GPT, and the rest — with JSON-first output that maps cleanly to scripts, CI pipelines, and agent runtimes.
# Install
curl -fsSL https://runapi.ai/cli/install.sh | sh
# Authenticate
runapi login
# Generate an image
runapi nano-banana generate \
--input '{"prompt":"a hummingbird drinking espresso"}'
# Generate music (fire-and-forget, then poll)
TASK_ID=$(runapi suno generate --async \
--input '{"prompt":"indie rock about coding"}' | jq -r .id)
runapi wait "$TASK_ID" --service suno --action generateWhat Is RunAPI CLI?
RunAPI CLI is a single Go binary that calls 130+ AI models from 18 providers through one unified API. Instead of installing separate SDKs for Suno, Kling, Flux, Claude, and GPT, you run one command — runapi <service> <action> — and get structured JSON back. It handles authentication, async polling, and output formatting so your shell scripts, CI pipelines, and agent runtimes stay clean. Every request is billed per-call with no monthly minimum, and the CLI stores your API key locally so you authenticate once and run commands from any directory.
Three Ways to Install
curl (Linux / macOS)
Auto-detects OS and architecture, verifies SHA-256 checksum, writes to /usr/local/bin.
curl -fsSL https://runapi.ai/cli/install.sh | shHomebrew
Official Homebrew tap. Auto-updates with brew upgrade.
brew install runapi-ai/tap/runapiGo Source Build
Build from source with Go 1.22+.
go install github.com/runapi-ai/cli/cmd/runapi@latestStart Building in Minutes
Authenticate, then generate.
# Browser-based login (laptop)
runapi login
# Or import a token (CI / headless)
printf '%s' "$RUNAPI_API_KEY" | runapi auth import-token --token -
# Verify
runapi auth statusrunapi nano-banana generate \
--input '{"prompt":"a hummingbird drinking espresso","aspect_ratio":"1:1"}'
# { "id": "task_abc123", "status": "completed", "output": { ... } }# Fire-and-forget for long jobs
TASK_ID=$(runapi suno generate --async \
--input '{"prompt":"upbeat indie rock about coding"}' | jq -r .id)
# Poll until completion
runapi wait "$TASK_ID" --service suno --action generateBuilt for Developers
JSON-First Output
Every command returns structured JSON — pipe to jq, feed to scripts, parse in any language.
Async + Polling
Fire-and-forget with --async, then poll with runapi wait. Built for long-running generation tasks.
Agent Runtime
Install CLI skills into Claude Code, Codex, Gemini CLI, and other agent runtimes with one command.
What Can You Generate?
Image Generation
Flux, Nano Banana, Recraft, Ideogram, Seedream, GPT Image, Imagen 4 — generate and edit images from text prompts.
Browse models →Video Generation
Kling, Veo, Runway, Luma, Hailuo, Wan, Seedance — text-to-video generation with async polling built in.
Browse models →Music & Audio
Suno for music, ElevenLabs for text-to-speech and voice cloning, InfiniteTalk for talking avatars.
Browse models →LLM Chat & Completion
Claude, GPT, Gemini, DeepSeek, Grok, Qwen — streaming chat completions via the OpenAI-compatible endpoint.
Browse models →Frequently Asked Questions
Three options: curl -fsSL https://runapi.ai/cli/install.sh | sh (auto-detects OS and arch), brew install runapi-ai/tap/runapi (Homebrew), or go install github.com/runapi-ai/cli/cmd/runapi@latest (Go source build).
macOS and Linux on both amd64 and arm64 architectures. The curl install script auto-detects your OS and CPU, verifies a SHA-256 checksum, and writes the binary to /usr/local/bin.
Yes. Use the curl installer in your CI/CD pipeline and set RUNAPI_API_KEY as an environment variable. The CLI supports headless authentication via runapi auth import-token for non-interactive environments.
Over 130 models across 18 providers: Suno for music, Kling and Veo for video, Flux and Nano Banana for images, ElevenLabs for audio, and Claude, GPT, Gemini, DeepSeek for LLMs.
Use the --async flag to fire-and-forget. The CLI returns a task ID immediately. Then use runapi wait TASK_ID to poll until completion. JSON output makes it easy to extract task IDs with jq.
The CLI itself is free and open source. You pay only for API calls on a per-request basis with no monthly minimum. Image generation starts at $0.02 per image, video generation from $0.10 per clip, and LLM calls are billed per token at roughly 50% below official provider pricing.
Yes. RunAPI publishes installable model skills for Claude Code, Codex, Gemini CLI, and other MCP-compatible agent runtimes. Run runapi skill install to add a model skill, then the agent can call RunAPI models as native tools without manual API configuration.
Explore More
MCP Server
Connect Claude Code, Cursor, and other MCP-compatible agents to RunAPI models.
SDKs
Python, Node.js, PHP, Java, Ruby, and Go SDKs for programmatic integration.
Skills
Install pre-built model skills into your agent runtime.
Model Catalog
Browse all 130+ models with pricing, parameters, and code samples.
Ready to run AI models from your terminal?
Install the CLI and start generating in minutes.