CLI

Install and use the Atlas Cloud CLI to chat with models, inspect schemas, and generate images or videos from your terminal.

The Atlas Cloud CLI provides the atlas command for accessing Atlas Cloud from a terminal or automation script. It supports authentication, LLM chat, model schema inspection, image generation, video generation, and prediction polling.

GitHub repository: AtlasCloudAI/cli

Install

Homebrew

brew install AtlasCloudAI/tap/atlascloud

The Homebrew formula is named atlascloud, but the installed command is atlas.

npm

npm install -g atlascloud-cli

The npm package is a lightweight wrapper that downloads the matching prebuilt release binary for supported platforms.

Windows PowerShell

irm https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.ps1 | iex

The PowerShell installer downloads the matching Windows release zip, verifies it against checksums.txt, installs atlas.exe, and adds it to the user PATH by default.

macOS / Linux Shell Installer

curl -fsSL https://raw.githubusercontent.com/AtlasCloudAI/cli/main/install.sh | sh

The shell installer supports macOS and Linux. For manual downloads, use GitHub Releases.

Verify

atlas version

Authenticate

Create an API key in the Atlas Cloud Console, then log in:

atlas auth login

For CI and non-interactive environments, pass the key with --token:

atlas auth login --token "$ATLASCLOUD_API_KEY"

Check local login state without calling the account API:

atlas auth status

Account and billing commands are currently marked as upcoming in the CLI. Use the console for account balance and billing until those endpoints are available.

Chat

Use atlas chat for OpenAI-compatible chat models:

atlas chat "Explain UUID v7 in two sentences" \
  --model deepseek-ai/DeepSeek-V3-0324

Pipe input from another command:

cat error.log | atlas chat "Find the root cause and suggest a fix"

When stdout is not a TTY, the CLI automatically uses JSON output. You can also force JSON:

atlas chat "say only OK" \
  --model deepseek-ai/DeepSeek-V3-0324 \
  --json | jq -r '.choices[0].message.content'

Models

List and search models:

atlas models list --json
atlas models search deepseek --json

Inspect a model schema:

atlas models get deepseek-ai/DeepSeek-V3-0324 --json
atlas models get google/nano-banana-2/text-to-image --json
atlas models get google/veo3.1/image-to-video --json
atlas models get bytedance/seedance-2.0-fast/image-to-video --json
atlas models get alibaba/wan-2.7/image-to-video --json

Current models list and models search results are strongest for chat models. Some image and video models are available through direct IDs even before they appear in the list. For media models, use the model documentation or a known model ID with atlas models get.

Generate images

Generate an image and wait for completion:

atlas generate image google/nano-banana-2/text-to-image \
  -p "a tiny cat"

Return immediately with a prediction ID:

atlas generate image google/nano-banana-2/text-to-image \
  -p "a tiny cat" \
  --no-wait \
  --json

Check or wait for an async prediction:

atlas generate get <prediction_id> --json
atlas generate wait <prediction_id> --json --no-download

By default, atlas generate image waits for completion and downloads outputs into the current directory. Use --no-download when an agent or script only needs output URLs.

Generate videos

Image-to-video models require an input image URL or local file:

atlas generate video google/veo3.1/image-to-video \
  -p "A cinematic camera push-in" \
  --image "https://example.com/input.png" \
  --resolution 1080p \
  --duration 8 \
  --no-wait \
  --json

Use atlas generate wait <prediction_id> to poll the task later.

Video model schemas vary. Common fields such as --image, --images, --end-image, --video, --audio, --resolution, --size, and --duration have first-class flags. For new or vendor-specific fields, pass exact schema fields with --params-json or repeated --param key=value:

atlas generate video google/veo3.1/image-to-video \
  --params-json '{"prompt":"A cinematic camera push-in","image":"https://example.com/input.png","resolution":"1080p","duration":8}' \
  --no-wait \
  --json

Agent-friendly usage

For scripts and AI agents, prefer stable machine-readable output:

atlas models get google/nano-banana-2/text-to-image --json

atlas generate image google/nano-banana-2/text-to-image \
  -p "a tiny cat" \
  --no-wait \
  --json

atlas generate video google/veo3.1/image-to-video \
  --params-json '{"prompt":"A cinematic camera push-in","image":"https://example.com/input.png","resolution":"1080p","duration":8}' \
  --no-wait \
  --json

atlas generate wait <prediction_id> \
  --json \
  --no-download

Recommended flags:

Flag	Use
`--json`	Force JSON output for parsing
`--no-wait`	Start a generation job and return the prediction ID immediately
`--no-download`	Print result URLs without writing files
`--quiet`	Suppress progress text in automation
`--params-json`	Pass an exact model input object from the model schema
`--param`	Add or override one model-specific field

Environment variables

Variable	Description
`ATLAS_API_BASE`	Override the API base URL. A trailing `/v1` is normalized automatically.
`ATLAS_TOKEN_FILE`	Store the CLI token in a custom file, useful for multiple accounts or CI.
`NO_COLOR`	Disable color output.

Known limits

atlas account and atlas auth whoami are upcoming until the server-side account endpoints are available.
atlas models list --type image and atlas models list --type video may be incomplete while /v1/models is still chat-model focused. Use atlas models get MODEL_ID --json for known media model IDs.
install.sh supports macOS and Linux. On Windows, use the PowerShell installer. Scoop/Winget can be added later as package-manager channels.

On This Page