Freepik is now Magnific

    Create images and video in ChatGPT and Claude

    Every Magnific model and tool, inside the chat you already use. Connect once and ask for the shot.

    One chat, from shoot to screen

    Real asks, real results. Every image, clip and track in this window came out of Magnific MCP.

    Bridge shot, 3 kits

    Magnific MCP
    Two women walking across a bridge, shot from a low fisheye angle

    Here's the bridge frame. Same shot, same two of them: dress them in 3 different kits.

    Thinking

    Create with Magnific in ChatGPT

    Ready-made asks for what people make most with the MCP. Pick one, attach your file and run it in your chat.

    • Campaign shots from one product photo

      Four images that keep your product exact: macro, lifestyle, studio and a splash.

    • Animate a still

      One frame becomes a 5-second shot that keeps the character and the light.

    • Music
      Effects

      Score and sound a clip

      Music to the clip's pace and an effect on every hit, laid under the picture.

    • Voice

      Voiceover for a scene

      Narration in the voice and tone you pick, timed to the shot.

    • Cut out a whole cast

      Every person on a clean transparent background, ready to drop anywhere.

    • Source video
      New cast

      Recast a video

      Put someone new on screen and keep every move, cut and camera angle.

    The models behind it

    Your agent calls the model that fits the job, or the one you name. Each one has its own page.

    One plan. Every chat.

    The MCP comes with every paid plan. Each ask spends credits from your plan, in ChatGPT, Claude or any chat you connect.

    Try a recipe: ask your assistant to swap the character in a video with Magnific Illusion

    FAQs

    • Where does Magnific MCP work?
      Claude (Desktop, Code, CLI), ChatGPT, Cursor, Codex, VS Code, Gemini, and Hermes.
    • What can I make with the Magnific MCP?
      Everything Magnific makes, from your chat. Ask in plain words and your agent picks the tool, runs it and brings the result back.
      • Images: Nano Banana, Seedream, GPT Image and more. Start from a prompt or from your own photo and keep the product, face or scene exact.
      • Video with sound: Kling, Veo and Seedance. Animate a still or describe the shot, the camera and the light.
      • Upscale and retouch: Upscale, relight, remove backgrounds and fix details without starting over.
      • Voice and music: Narration in any voice, music to your tempo, sound effects and dubbing.
      • Your library: Search everything you've made, and Magnific's stock, without leaving the chat.
      • Workflows: Save a pipeline in Spaces once. Run it again from any chat.
    • Do I need an API key?
      No. Sign in with your Magnific account when your client asks, approve the connection, and you're in. No keys to copy, store or rotate.
    • Which models can I call?
      Nano Banana 2, GPT 2, Seedream 5, Google Imagen 4, Recraft V4, Seedance 2.0, Kling 3.0 Omni, Happy Horse, or Google Veo 3.1, among others. Catalog updated daily.
    • Can I generate videos with audio?
      Yes. Pick models like Seedance 2.0, Kling 3.0, or Google Veo 3.1.
    • Can I run bulk operations?
      Yes. Batch image generation, character creation, and video concatenation are available now. Folder-level upscale is coming soon.
    • Can I edit Spaces from the chat?
      Yes. Tell the chat what to change and it updates the Space for you.
    • Does my plan cover the MCP?
      Yes. Magnific MCP is available on all paid plans. Keep in mind that MCP actions always consume credits. Even if your plan includes credit-free generations on certain models, those models still consume credits when accessed via MCP.
    • Can I see the cost before I generate?
      Yes. Ask what it will cost. The MCP estimates the credits for any generation before it runs, and the estimate spends nothing. It also tells your agent to check the cost before using credits.
    • Is it listed in Claude's official directory?
      Custom connector for now. Official directory listing in progress.
    • Do you train AI models on my data?
      No. We never train on your data: not your prompts, not your creations. Everything you make stays in your Magnific workspace.
    • How can I know which model is being used?
      The server automatically chooses the best model based on the context—this is auto mode. The MCP doesn't explicitly return which model it chose in auto mode, so there's no way to know for sure without checking the resulting creation afterward.
      For full control, you can ask it to use a specific model, for example, "generate an image with Google Nano Banana 2" or "use Recraft V4.1."
    • What's the difference between the MCP and the API?
      The API is for developers who want Magnific wired directly into their own products and pipelines, with full control in their own code. The MCP lets AI agents like Claude, ChatGPT, or Cursor use Magnific on their own, in plain language, no code required.
    • Besides generating content, what else can I do with the MCP?
      • Check your account balance: You can ask how many credits you have available at any time, without spending a single one.
      • Browse and display your existing creations: You can search your generation history and render any past image, video, or audio directly in the conversation.
      • Combine with other connectors: The MCP works alongside other integrations in the same conversation. For example, you could pull up a Magnific image and then share it via Slack or drop it into a Gmail draft.
      None of these actions trigger any credit consumption. Credits only apply when you actually generate or transform content (image generation, upscaling, expanding, retouching, etc.).

    Bring Magnific to your next chat

    Connect once. Every image, video and workflow is one ask away.

    Connect your agent