Global background decoration
Whisk AI MCP · Beta

Stay in the chat.
Make something.

Connect Whisk AI to ChatGPT and Claude. Generate images and videos with multiple image and video models using your existing credits.

No API keys · Your existing credits · One media library

An example workflow

You

Turn this product shot into a vertical video. Show me the credit estimate first.

Whisk AI

Find your reference image
Read Kling 3.0 settings
Estimate credits before submitting

Product video

Kling 3.0 · 9:16 · 5s

Estimate

Your direction. Your tools. One conversation.

01 / Connect

Meet your next creative tool.

Open ChatGPT

The button copies the URL and opens setup. Paste the URL, add the connection, and authorize Whisk AI to finish.

  1. 1

    Click “Copy URL & open ChatGPT”, then choose Add custom MCP server in Plugins.

  2. 2

    Name it Whisk AI, paste the server URL below, and select OAuth.

  3. 3

    Create and install the plugin. Sign in to Whisk AI and approve access when prompted.

Use your Whisk AI account. No API key needed. Organization accounts may require an administrator to enable custom connectors.

02

Start with what is yours.

Upload a product photo or choose an image from your library to guide the next edit or video.

03

Know the cost first.

Ask your agent to check model settings, your balance, and the credit estimate before creating.

04

Keep every next step close.

Completed results are saved to Whisk AI, ready to revisit after the conversation ends.

From a still to a story.

Website upload

Image generation & editing

Nano Banana 2, Nano Banana Pro, Nano Banana 2.1, Nano Banana Lite, GPT Image 2.5, and Seedream 4.5

Resolution and reference limits vary by model

Video generation

Kling 3.0, Grok Video, Grok Video 1.5, Kling 4.0, Seedance 2.0, Seedance 2.0 Mini, Seedance 2.5, Sora 2, and Sora 2 Pro

Duration, resolution and reference modes vary by model

Before you connect.

Do I need a separate MCP subscription?

MCP uses your existing Whisk AI credits. Cost depends on the selected model and settings; your agent can check an estimate before submitting.

Which models are available?

Image generation and editing: Nano Banana 2, Nano Banana Pro, Nano Banana 2.1, Nano Banana Lite, GPT Image 2.5, and Seedream 4.5. Video: Kling 3.0, Grok Video, Grok Video 1.5, Kling 4.0, Seedance 2.0, Seedance 2.0 Mini, Seedance 2.5, Sora 2, and Sora 2 Pro. Read each model’s parameters before generating.

How do I use a photo from my computer?

Chat image import is being tested with ChatGPT: attach a JPEG, PNG or WebP (up to 8 MB) and ask your agent to save it as a reference, without copying an asset ID. Claude attachment transfer is still unverified; use the website upload or select a library image for now. Video reference files are not supported yet.

What if a result takes a while?

Generations run as background tasks. Your agent can check progress or share a link to your workspace. A status timeout does not mean generation failed; avoid submitting the same task again.

How do I disconnect?

Open Connected apps while signed in and revoke the connection. This stops future access; generations you already submitted can still finish.