NOW LIVESeedance 2.5 is now live on I2V.aiTry it now
i2v.aii2v.aii2v.ai
  • Studio
  • Compare
  • Market
  • Pricing
  • Docs
i2v.aii2v.aii2v.ai

I2V AI — image and video creation, with API access

  • GitHub

Product

  • Studio
  • Image Generation
  • Video Generation
  • Models
  • Pricing
  • Video API
  • Image API

Resources

  • Service status
  • CLI & Skills
  • MCP
  • llms.txt
  • Scenes
  • Works
  • Prompts
  • Image to Prompt
  • Batch Image to Prompt

Image Models

  • Seedream 5.0
  • Nano Banana Pro
  • Nano Banana Flash
  • Nano Banana 2

Video Models

  • Google Veo 3.1
  • Google Veo 3.1 Pro
  • Seedance 2.0
  • Kling v3.0
  • Kling v3.0 Pro

Company & Legal

  • About
  • Contact
  • Privacy Policy
  • Terms of Service
  • Refund Policy
  • Media retention

This website is an independent third-party platform. We are not affiliated with, endorsed by, or officially representing any AI model providers referenced on this site. All trademarks and brand names belong to their respective owners.

© 2026 I2V AI. All rights reserved. Operated by CLEANCUT RAZOD LLC

support@i2v.aiprivacy@i2v.aiDiscord

i2v.ai CLI

Generate images and video from your terminal

Use the i2v.ai CLI to script generation, upload local files, get a USD quote, and save results in your project. Use Skills for guided creation or API integration.

Get an API keyDownload CLI
npx skills add https://i2v.ai --skill i2v-models

Install the media workflow in a supported agent. Set I2V_API_KEY if the agent will use the bundled CLI.

First image from your terminal

Set your key in the terminal and open your project directory. Use plan to check the model parameters and account quote before submitting.

# Set the key in the shell that starts your agent
export I2V_API_KEY="YOUR_API_KEY"

node i2v-cli.mjs doc gpt-image-2
node i2v-cli.mjs plan gpt-image-2 --input '{"prompt":"A ceramic mug on an oak table"}'

# Use your approved budget and the expectedRevision from plan
BUDGET_USD="YOUR_APPROVED_USD_BUDGET"
REVISION="MODEL_REVISION_FROM_PLAN"
node i2v-cli.mjs run gpt-image-2 --input '{"prompt":"A ceramic mug on an oak table"}' \
  --max-cost-usd "$BUDGET_USD" --expected-revision "$REVISION" --no-wait
node i2v-cli.mjs status TASK_ID --wait 120 --out ./mug.png

Set --max-cost-usd to a budget you approve. Read the task ID in the receipt, then use status to save the file. If the task ID is lost, use recover with the original submission key and credential.

Setup on Windows
curl.exe -fsSLo i2v-cli.mjs https://i2v.ai/cli/i2v-cli.mjs
$env:I2V_API_KEY = "YOUR_API_KEY"
node .\i2v-cli.mjs list --type image
Browse models

Connect your agent

Remote Streamable HTTP

Use the remote URL in an MCP client that supports Streamable HTTP. Choose OAuth where supported, or configure your API key as a Bearer token in the client.

Client setup and connection test
https://i2v.ai/api/mcp

Local stdio with file access

For a client that launches a local MCP server, download the CLI and configure the command below. Set I2V_API_KEY in its environment. Set I2V_WORKSPACE_DIR to enable uploads and downloads in that project.

node /absolute/path/i2v-cli.mjs mcp
Example desktop client configuration
{
  "mcpServers": {
    "i2v": {
      "command": "node",
      "args": [
        "/absolute/path/i2v-cli.mjs",
        "mcp"
      ]
    }
  }
}

Command reference

list [--type image|video] [--q QUERY]
Discover published models.
doc MODEL [--schema] [--locale zh]
Read Markdown documentation or the model schema.
price [QUERY]
Show public catalog USD rates.
balance
Read your available and reserved USD.
usage [--range 30d]
Read settled charges, refunds and net USD usage for a UTC date range.
recommend [QUERY] [--task TYPE]
Find model candidates for a task and its media constraints.
plan MODEL --input-file FILE
Validate inputs and preview the account quote without submitting.
quote MODEL --input-file FILE
Quote the actual inputs on your account.
upload FILE
Upload and verify a local reference file.
run MODEL --input-file FILE --max-cost-usd USD
Submit with a USD ceiling and wait or return a receipt.
status TASK_ID [--wait SECONDS] [--out FILE]
Resume polling and download an existing result.
recover IDEMPOTENCY_KEY
Find an original submission by its idempotency key.
doctor
Check the installed CLI and Skill versions, and inspect the configured MCP connection.
mcp
Start the local stdio MCP server.
Can I browse without a key?

Yes. list, doc and price read the public catalog. Quote, balance, upload, generation and task status use your account.

What happens if the agent times out?

Use status with the task ID. If you only have the original idempotency key, use recover with the same credential.

Where can MCP save files?

Local stdio can save results inside the configured workspace. A remote MCP connection returns result URLs.

How do I control spending?

Set a task budget with --max-cost-usd and manage cumulative spending in API Key settings. With OAuth, grant write access only to clients that should submit tasks.