Everypixel CLI

Official command-line interface for the Everypixel API. Generate and edit images and videos, chat with GLM-5.3, work with audio, run visual analysis, and automate Everypixel workflows from a terminal or an MCP client.

Commands return readable terminal output by default and JSON for scripts. The package includes a stdio MCP server for agent integrations.

API documentation · GitHub · Issues · Changelog

Quick start
Images, video, audio, chat, analysis
Tasks and downloads
JSON output and generic API calls
Authentication and profiles
MCP server
Global options, exit codes, troubleshooting

Installation

Everypixel CLI requires Python 3.10 or newer.

pip install everypixel-cli

The package installs two equivalent commands:

everypixel --help
epx --help

Quick Start

Create API credentials in your Everypixel Labs account. Run the interactive setup and enter the client ID and secret when prompted:

everypixel auth configure
everypixel auth check

Generate an image and save it locally:

everypixel image generate "editorial product photo on a white background" \
  --model flux3 --download ./outputs

Media commands wait for completion by default. See Tasks and downloads for background jobs and JSON output for scripts.

Use --help for command syntax. To list image generation models, select their enum from OpenAPI:

everypixel image generate --help
everypixel docs openapi --jq .components.schemas.ImageGenerateModelEnum.enum

Examples use Bash syntax. Examples that extract IDs with jq require the separate jq command.

Authentication

Everypixel uses HTTP Basic Auth with the client ID as the username and the client secret as the password. You can save credentials with auth configure or provide them through environment variables:

export EVERYPIXEL_CLIENT_ID="your-client-id"
export EVERYPIXEL_CLIENT_SECRET="your-client-secret"

Variable	Purpose
`EVERYPIXEL_CLIENT_ID`	API client ID
`EVERYPIXEL_CLIENT_SECRET`	API client secret
`EVERYPIXEL_BASE_URL`	Override the API base URL
`EVERYPIXEL_PROFILE`	Select a saved configuration profile
`NO_COLOR`	Disable colored terminal output when set

Explicit command-line options take priority over environment variables, and environment variables take priority over saved configuration.

Useful authentication commands:

Command	Description
`everypixel auth configure`	Save credentials for the current profile
`everypixel auth check`	Verify the configured credentials
`everypixel auth whoami`	Show the active profile and credential source
`everypixel auth logout`	Remove credentials from the current profile

Secrets are stored in the operating system keyring when one is available. The CLI falls back to its user configuration file when a keyring is unavailable; everypixel config list masks saved secrets in its output.

Commands

Images

Command	Description
`everypixel image generate PROMPT`	Generate an image from a text prompt
`everypixel image edit PROMPT --image FILE`	Edit one or more images
`everypixel image vectorize FILE`	Convert an image to SVG
`everypixel image upscale`	Upscale an image, URL, or existing task result
`everypixel image angles --image FILE`	Change the camera angle of an image
`everypixel image colors --image FILE --reference FILE`	Transfer colors from a reference image

Examples:

everypixel image edit \
  "replace the background with a clean studio wall" \
  --image ./input.png \
  --model qwen \
  --download ./outputs

everypixel image upscale --image ./input.png --download ./outputs

everypixel image generate "a mountain icon" \
  --model recraftv4_1_vector \
  --controls '{"colors":[{"rgb":[20,100,180],"weight":1}]}' \
  --download ./outputs

everypixel image vectorize ./input.png --download ./outputs

everypixel image angles \
  --image ./input.png \
  --azimuth right \
  --elevation eye_level \
  --distance medium \
  --download ./outputs

Recraft V4.1 vector models (recraftv4_1_vector, recraftv4_1_pro_vector) produce SVG files and accept optional JSON palette --controls. Use --size to select their aspect ratio. Raster models also accept --resolution.

--lora-url is supported only by zimage.

GPT Image models are gpt-image-2 and gpt-image-2.5-sunburst. Both generation and editing accept --quality low|medium|high (default: medium); Sunburst also supports xhigh and max. Use --size square with --resolution 3k.

everypixel image generate "a detailed botanical illustration" \
  --model gpt-image-2.5-sunburst --quality high --download ./outputs

Other image models include flux3, gemini-nano-banana-2.1 (Nano Banana 2.1), gemini-3.1-flash-lite (Nano Banana 2 Lite), and seedream-5-flash. Nano Banana 2 Lite uses --resolution 1k. Repeat --image to provide multiple inputs to an editing model.

Video

Command	Description
`everypixel video generate PROMPT`	Generate video from text
`everypixel video from-image PROMPT --image FILE`	Animate an image
`everypixel video first-last PROMPT`	Generate video between first and last frames
`everypixel video edit`	Edit video with text and media references
`everypixel video upscale`	Upscale a video, URL, or existing task result

Animate a local image or continue a video:

everypixel video from-image "subtle camera movement and natural wind" \
  --image ./first-frame.png --model ltx23 --download ./outputs

everypixel video edit "continue the scene with a slow camera pan" \
  --model flux3 --video ./input.mp4 --download ./outputs

Generate a video with Gemini Omni 1.1 Flash, then edit the completed task:

task_id="$(
  everypixel video generate "a slow aerial shot over mountain lakes" \
    --model gemini-omni-1.1-flash --output-json | jq -er '.task_id'
)"

everypixel video edit "change the scene to sunset" \
  --model gemini-omni-1.1-flash --task-id "$task_id" --download ./outputs

Gemini Omni editing requires an owned Gemini Omni source task. Other editing models accept --task-id as an alternative to --video.

Input options depend on the model:

first-last accepts --image and --last-image for models that support frame pairs.
MiniMax and WAN 3.0 accept repeatable reference options, separately from frame inputs. Use --reference-image, --reference-video, and, for MiniMax, --reference-audio.
Gemini Omni generation accepts frame images or --reference-image, separately.
Grok Imagine 1.5 (grok-imagine-1.5) requires --image and supports 1080p.
Aleph 2 accepts repeatable --keyframe JSON; inspect video edit --help for its fields.

See the API documentation for each model's accepted inputs, resolutions, and durations.

For upscaling, use seedvr2 (the default) or a Topaz model such as topaz-prob-4:

everypixel video upscale --video ./input.mp4 \
  --model topaz-prob-4 --resolution 4k --download ./outputs

Lipsync and Audio

Command	Description
`everypixel lipsync video`	Synchronize a video with an audio track
`everypixel lipsync image`	Create a lipsync video from a still image and audio
`everypixel audio transcribe`	Transcribe speech from an audio file
`everypixel audio tts-create`	Create speech with a Qwen preset or saved ElevenLabs voice
`everypixel audio tts-clone`	Create Qwen speech from an audio sample or saved voice
`everypixel audio tts-design`	Design voice previews from a description
`everypixel audio tts-voice`	Save audio samples or a design preview as a voice
`everypixel audio tts-voices`	List presets and saved voices
`everypixel audio tts-delete`	Delete an owned saved voice

Synchronize video, transcribe audio, or synthesize speech with a Qwen preset:

everypixel lipsync video --video ./input.mp4 --audio ./voice.mp3 \
  --download ./outputs

everypixel audio transcribe --audio ./speech.mp3 --language English \
  --download ./outputs

everypixel audio tts-create --model qwen3 --text "Hello world" \
  --speaker Ryan --style Warm --language English --download ./outputs

--speaker, --style, --language, and --seed are Qwen controls; --character applies only to Qwen voice design. ElevenLabs synthesis uses --voice-id instead of preset controls. tts-clone and tts-design default to Qwen; tts-create lets the API choose when --model is omitted.

Design an ElevenLabs preview, save the first preview as a voice, and use it:

preview_id="$(
  everypixel audio tts-design --model eleven_ttv_v3 \
    --prompt "A warm, calm narrator with clear diction and a relaxed pace." \
    --text "Welcome to our guided tour. Today we will explore the stories behind these photographs and the people who made them." \
    --output-json | jq -er '.result.previews[0].preview_id'
)"

voice_id="$(
  everypixel audio tts-voice --name "Narrator" --preview-id "$preview_id" \
    --output-json | jq -er '.voice.id'
)"

everypixel audio tts-create --model eleven_v3 --voice-id "$voice_id" \
  --text "Hello world" --download ./outputs

To save your own samples, use tts-voice --name "My voice" --provider elevenlabs --audio ./sample.wav; repeat --audio for more samples. Qwen uses --provider labs and one sample. Saved Qwen voices work with tts-clone --voice-id.

List saved ElevenLabs voices with everypixel audio tts-voices --provider elevenlabs. See the API documentation for model choices and text constraints.

Chat

everypixel chat "Explain how a rainbow forms."
everypixel chat "Review this idea." --system "Be concise." --stream
everypixel chat "Solve this problem." --reasoning-effort high --show-reasoning
everypixel chat --input-file ./conversation.json --output-json

Chat uses the same credentials as other commands and requires the client's chat scope. The default model is glm-5.3. Each invocation sends one request; use --input-file for conversation history, function tools, or other API options:

{
  "messages": [
    {"role": "user", "content": "What causes rainbows?"},
    {"role": "assistant", "content": "Sunlight refracts and reflects in water droplets."},
    {"role": "user", "content": "Why do the colors separate?"}
  ],
  "max_completion_tokens": 4096
}

A positional prompt appends a user message to the supplied history. CLI options override matching fields in the file. --stream displays text as it arrives; JSON output returns one assembled response, including reasoning, tool calls, and usage. Function calls are returned for the caller to handle. The MCP chat tool accepts the same structured request and returns a complete response. Use --request-timeout to change the chat HTTP timeout (default: 240 seconds).

Visual Analysis

Analysis commands accept HTTP(S) URLs and local files. Local files are uploaded automatically when the endpoint supports them.

Command	Description
`everypixel keywords --image FILE`	Extract image keywords
`everypixel quality --image FILE`	Score technical image quality
`everypixel quality-ugc --image FILE`	Score user-generated image quality
`everypixel faces --image FILE`	Detect faces in an image
`everypixel captioning --image FILE`	Generate an image caption
`everypixel video-keywords --video FILE`	Extract video keywords

everypixel keywords \
  --image ./image.jpg \
  --lang en \
  --num-keywords 10 \
  --colors \
  --output-json

everypixel captioning --image ./image.jpg --output-json

Async Tasks and Downloads

Generation and media-processing commands wait for completion by default. Add --no-wait to return the newly created task immediately.

everypixel image generate "minimalist poster" --no-wait --output-json

Use status for one status request and wait to poll until the task reaches a terminal state:

everypixel status TASK_ID
everypixel wait TASK_ID

Passing --download always waits for a successful result before saving files:

everypixel wait TASK_ID --download ./outputs

elapsed_sec is total local waiting time; queue_sec estimates time until the first observed STARTED response.

API costs are USD strings: estimated_cost for pending tasks and billed_cost for successful results and chat. Missing costs are omitted; "0" means no charge.

Generic OpenAPI Runner

everypixel run calls an API operation by its OpenAPI name or /v1/... path. Values passed with --input are parsed as JSON when possible. The CLI validates named operations against the selected OpenAPI schema before sending a request.

everypixel run image_generate \
  --prompt "product photo" \
  --input model=flux2 \
  --input image_size=square \
  --dry-run

everypixel run image_generate --help-schema --output-json

everypixel run /v1/image_generate \
  --method POST \
  --input-file ./payload.json

--help-schema shows the operation definition. Referenced component schemas are included in the full document returned by docs openapi.

The schema loader tries the live API first, then the local cache, and finally the schema bundled with the package.

everypixel docs open
everypixel docs openapi --output-json
everypixel docs refresh

JSON Output

Use --output-json (or -j) for scripts and --jq to select part of the result:

everypixel keywords --image ./image.jpg --output-json
everypixel keywords --image ./image.jpg --output-json --jq .keywords

Handled successes and errors are written as one machine-readable JSON stream to stdout. Human-readable errors are written to stderr. Add --debug when an unexpected internal error needs sanitized stack-frame locations.

Global Options

Global options must appear before the command they apply to.

Option	Description
`--base-url URL`	Override the API base URL for one invocation
`--profile NAME`	Use a saved configuration profile
`--output human\|json`	Select the output format
`--output-json`, `-j`	Shortcut for JSON output
`--jq EXPRESSION`	Filter JSON output
`--wait`, `--no-wait`	Control async task polling
`--timeout SECONDS`	Set the polling timeout (default: 30 minutes)
`--poll-interval SECONDS`	Set the delay between status requests
`--download DIRECTORY`, `-d DIRECTORY`	Download completed result files
`--no-color`	Disable colored terminal output
`--debug`	Include sanitized diagnostic locations for internal errors

Operation commands also accept common options such as --download, --no-wait, and --output-json after the command. Run everypixel COMMAND --help for the exact syntax.

Configuration Profiles

Profiles keep credentials and settings separate for different accounts or projects. Most users can stay on the default profile.

everypixel config profile create project-a
everypixel config profile use project-a
everypixel auth configure
everypixel config list

Select a profile for one invocation without changing the active profile:

everypixel --profile project-a auth check

MCP Server

Everypixel CLI includes an MCP server for Claude Desktop, Claude Code, Cursor, VS Code, and other clients that support MCP over stdio:

everypixel mcp serve

Example client configuration:

{
  "mcpServers": {
    "everypixel": {
      "command": "everypixel",
      "args": ["mcp", "serve"],
      "env": {
        "EVERYPIXEL_CLIENT_ID": "your-client-id",
        "EVERYPIXEL_CLIENT_SECRET": "your-client-secret"
      }
    }
  }
}

The MCP client launches the command as a subprocess; no HTTP port is opened. Local media and download paths use the subprocess working directory and filesystem permissions.

Shell Completion

Install completion for the current shell:

everypixel --install-completion

Or print the completion script for manual setup:

everypixel --show-completion

Exit Codes

Code	Meaning
`0`	Success
`1`	General error
`2`	Local argument or validation error
`3`	Configuration or authentication error
`4`	API, remote service, or task failure
`5`	File read/write or result download error
`6`	Result processing, jq, or serialization error

Troubleshooting

Authentication failure: run everypixel auth check, then configure the credentials again if needed.
Wait timeout: increase the global timeout, for example everypixel --timeout 7200 wait TASK_ID.
Invalid media: pass an existing local file, an HTTP(S) URL, or a supported data URI.
File not found: use an absolute path or run the command from the directory containing the file.
Download failure: verify that the output directory is writable and retry with everypixel wait TASK_ID --download ./outputs.
No color: pass --no-color or set NO_COLOR=1.

License

MIT

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
src/everypixel_cli		src/everypixel_cli
tests		tests
.gitattributes		.gitattributes
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
LICENSE		LICENSE
README.md		README.md
pyproject.toml		pyproject.toml

Navigation Menu

Repository navigation

About

Repository files navigation

Everypixel CLI

Installation

Quick Start

Authentication

Commands

Images

Video

Lipsync and Audio

Chat

Visual Analysis

Async Tasks and Downloads

Generic OpenAPI Runner

JSON Output

Global Options

Configuration Profiles

MCP Server

Shell Completion

Exit Codes

Troubleshooting

License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

Everypixel CLI

Installation

Quick Start

Authentication

Commands

Images

Video

Lipsync and Audio

Chat

Visual Analysis

Async Tasks and Downloads

Generic OpenAPI Runner

JSON Output

Global Options

Configuration Profiles

MCP Server

Shell Completion

Exit Codes

Troubleshooting

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages