Loading PowerTokens
Wait a moment...
Wait a moment...
PowerTokens CLI is a developer-first command line tool. Call DeepSeek, GLM, Qwen, Seedream, Seedance and more from a single terminal entry point — no need to maintain a separate integration for every model.
npm install -g powertokens-cliPowerTokens CLI collapses install, auth, model selection and result handling into the terminal experience developers already use every day.
One CLI experience across every model, so switching providers never means learning a new tool.
Chat, image, video and audio from DeepSeek, GLM, Qwen, Seedream, Seedance, Kling, MiniMax and more.
Built for Terminal, shell scripts and automation, instead of being locked inside a web playground.
Embed model calls into build steps, batch jobs and internal tooling with plain commands.
Switch models with a single flag instead of rebuilding a separate integration for each one.
Released under the MIT License, so you can install it, use it and build on top of it.
From installation to your first model call, the path stays as short as possible.
npm install -g powertokens-clipt login
pt whoamipt chat "What is TypeScript?"
pt chat -m deepseek-v4-pro "Write a poem"
echo "Explain recursion" | pt chat# Image
pt generate -m seedream-5-0-260128 -p "cyberpunk city" -o city.png
# Video
pt generate -m dreamina-seedance-2-0-260128 -p "a dog on the beach" -o dog.mp4
# Audio (TTS)
pt generate -m speech-2.6-hd -p "Hello from PowerTokens" -o hello.mp3pt login [key]Configure and store your PowerTokens API key.pt whoamiShow the current API key prefix, base URL and model count.pt logoutRemove the stored API key from your machine.pt chat [prompt]Chat with an LLM model. Reads from stdin when no prompt is given.pt generateGenerate images, video or audio from a prompt or reference media.pt modelsList available models, optionally filtered by -t chat|image|video|audio.pt configShow or set the API base URL and asset library base URL.The real value of a CLI is turning a model call into an ordinary command in your development process.
Try different models, prompts and parameters right inside your terminal.
Drop AI requests into existing shell, Node.js or Python workflows.
Run model calls inside batch jobs for bulk content generation and processing.
Make AI a step in your delivery pipeline and cut down manual work.
Run pt models to fetch the live list. A quick snapshot of the families you can reach:
Settings live in a single config file and can always be overridden by environment variables — handy for CI and shared machines.
POWERTOKENS_API_KEYAPI key (overrides the config file).
POWERTOKENS_BASE_URLAPI base URL (overrides the config file).
POWERTOKENS_ASSET_BASE_URLAsset library base URL, auto-derived from the base URL.
PT_DEBUGSet to 1 to print the raw request body and response to stderr.
pt models # all models
pt models -t image # image models only
pt models -t video # video models only{
"apiKey": "sk-...",
"baseUrl": "https://api.powertokens.ai",
"assetBaseUrl": "https://powertokens.ai/api"
}