MCP server · 19 tools

Generate video, images and voice from Claude, ChatGPT and Cursor

Add Klakix to your AI assistant once, then ask for a video in plain words. It picks a model, tells you the credit cost, starts the job and brings the finished file back to the chat.

Server URL

https://klakix.com/api/mcp
Example chat

You

An epic sunrise over Cappadocia: hundreds of colorful hot air balloons drift above the fairy chimneys, golden sunbeams cut through the morning haze, and a crane shot rises to reveal the valley. Rich detail, large-format film look.

Claude

generate_video · Seedance 2.5 · 16:9 · 720p · 5 sec

That's 147 credits for a 5-second 720p clip with Seedance 2.5. I started it, and here's the result:

An epic sunrise over Cappadocia: hundreds of colorful hot air balloons drift above the fairy chimneys, golden… Still from a video made with Seedance 2.5.
Seedance 2.5An epic sunrise over Cappadocia: hundreds of colorful hot air balloons drift above the fairy chimneys, golden sunbeams cut through the morning haze, and a crane shot rises to reveal the valley. Rich detail, large-format film look.
The prompt and the clip are a real Seedance 2.5 sample from our examples, made on Klakix. The chat around them shows how the same request looks in Claude; the credit cost is today's price for these settings.

19 tools

What you can make from chat

Your assistant reads each tool's description and picks the right one. You describe the result; it chooses the model, length and aspect ratio, or uses the ones you name.

Create

9
  • generate_videoUses credits

    Video from text, a photo or references (Veo, Kling, Seedance, Wan…).

  • generate_imageUses credits

    Images from text (Nano Banana 2, GPT Image 2.5, Muse).

  • generate_audioUses credits

    Voiceover, multi-voice dialogue, music and sound effects.

  • create_talking_videoUses credits

    Talking photos and UGC-style ad videos with lip sync.

  • apply_effectUses credits

    Ready-made video effects on a photo (hug, dance, transformations…).

  • transfer_motionUses credits

    Moves a character photo the way a reference video moves.

  • upscale_videoUses credits

    Upscales a video to 1080p or 4K, optionally to 60 fps.

  • create_explainer_videoUses credits

    Narrated, captioned explainer video from a topic (Motif).

  • run_templateUses credits

    One-step video or image from a ready-made template.

Files and results

4
  • upload_mediaNo credits

    Imports an image, video or audio from a URL or an earlier result.

  • get_generationNo credits

    A generation's status and, when done, its file links (can wait).

  • list_generationsNo credits

    Recent generations in your library.

  • cancel_generationNo credits

    Cancels a queued generation and refunds its credits.

Account and catalog

6
  • get_accountNo credits

    Credit balance, plan and concurrent generation limit.

  • list_modelsNo credits

    Models, what they can do and their credit prices.

  • get_modelNo credits

    Every setting, limit and example prompt for one model.

  • estimate_costNo credits

    Works out the credit cost before you generate.

  • list_voicesNo credits

    Voiceover voices with preview links.

  • list_templatesNo credits

    Ready-made templates and the fields they need.

Video models: Seedance 2.5, Seedance 2.0, Seedance 2.0 Mini, MiniMax H3 Max Turbo, MiniMax H3 Max, Google Gemini Omni Flash, Google Veo 3.1 Lite, Wan 3.0, Kling 4.0, and Kling 3.0. Image models: Muse Image, Nano Banana 2, and GPT Image 2.5. Talking videos: MiniMax H3 Max Lip Sync, HeyGen Avatar IV, and OmniHuman 1.5. Each one is described on AI video models.

7 ready-made workflows

A workflow is a step-by-step recipe the server hands to your assistant: it writes the script, picks the models and runs each step in order. If your app supports MCP prompts you can pick one from its menu (in Claude Code they appear as slash commands); otherwise just ask for it in your own words.

  1. Short ad video from a product photo

    product_ad
  2. Presenter-led UGC ad (script + talking video)

    ugc_ad
  3. Multi-shot short film from a script

    storyboard_film
  4. Bring a photo to life

    photo_to_video
  5. Vertical social video (music + captions)

    social_reel
  6. Voiceover, music and sound effects from a script

    voiceover_pack
  7. Several scenes with the same character

    character_consistency

Setup

Add it to your app

Pick your app. Most connect by logging in to your Klakix account, so there's nothing to install and no key to copy.

Claude

Chat app

claude.ai, the Claude desktop app and mobile app. The Free plan allows one custom connector; on Team and Enterprise an Owner adds it. Add it once and it works on every device.

  1. In Claude, open Customize > Connectors (on the web or desktop).
  2. Click + Add, then choose Add custom connector.
  3. Enter a name, paste the server URL below and click Continue.
  4. Keep the detected authentication settings, click Continue and then Add; when asked, log in to your account and approve access.
  5. In a new chat, click +, open Connectors, turn the connector on and ask "How many credits do I have?"

Connects by logging in to your account (OAuth); no key needed

Server URL

https://klakix.com/api/mcp

Already connected? See your connected apps, API keys and recent activity on the MCP page in the studio.

Pricing

How credits are charged

  • The same price as the web. Each generation uses credits from your Klakix account at the price the studio shows. There's no MCP fee and no separate plan.
  • Check before you spend. Ask “how much will this cost?” and the assistant calls estimate_cost before it generates.
  • Refunds. A failed generation is refunded automatically, and cancel_generation refunds a job that's still queued.
  • Before your first purchase only voiceover works. Video, image and talking-video tools become available once you buy credits; packs start at $5 for 75 credits, and credits never expire.
  • Generations at once: 3 without a plan, 4, 6, and 10 on the Starter, Growth and Agency plans.

Example costs

  • Kling 4.05 sec video, 720p79credits · $2.87–$5.27
  • Seedance 2.55 sec video, 720p147credits · $5.34–$9.80
  • Nano Banana 21 image, 1K8credits · $0.29–$0.53

The dollar range runs from the largest to the smallest credit pack. Full list on pricing.

Background

What is MCP?

MCP (Model Context Protocol) is an open standard that Anthropic introduced in November 2024 for connecting AI assistants to outside tools and data. A server describes the tools it offers; an MCP client such as Claude, ChatGPT or Cursor calls them when your request needs one.

Klakix's server is a remote one. You add its URL to your app, log in once, and the tools show up in your chat. Generation happens on Klakix's side, so your assistant doesn't need a video or image model of its own.

Comparison

Compared with Higgsfield MCP

Higgsfield also offers an MCP server for Claude and other assistants. Here's how the two compare on the points both publish.

Klakix MCPHiggsfield MCP
Server URLhttps://klakix.com/api/mcphttps://mcp.higgsfield.ai/mcp
Sign-inOAuth, or an API key for tools without OAuthOAuth, no API key needed
Plan neededNone: any credit pack (voiceover only before the first purchase)An active paid subscription
Credits in MCPThe same price as the web; check it first with estimate_cost; failed generations refundedEvery generation uses credits; unlimited models and free generations work only on higgsfield.ai
Documented appsClaude, ChatGPT, Claude Code, Cursor, VS Code, Devin Desktop (Windsurf), Codex, and Gemini CLIClaude, Cursor, ChatGPT, and Grok
PricesCredit packs in US dollars before tax, or local currency where availablePlan credits; prices exclude VAT and local taxes

Higgsfield details from its MCP page, help center and pricing page, checked October 15, 2026. Klakix isn't affiliated with Higgsfield.

Also in Klakix's MCP

  • Talking videos and UGC-style ads with lip sync (MiniMax H3 Max Lip Sync, HeyGen Avatar IV, and OmniHuman 1.5).
  • Animated explainer videos with voiceover and captions (Motif).
  • Ready-made templates that run in one step (run_template).
  • Voiceover, multi-voice dialogue, music and sound effects in the same account.

More on plans and prices: Higgsfield alternative.

Where Higgsfield's MCP goes further

  • More models, including its own Soul characters.
  • Voice cloning, voice changing and dubbing.
  • Background removal, reframing and building apps and games from chat.
  • A separate Bridge MCP that works inside Blender and After Effects.

Security

Permissions you control

  • Your password stays with us

    You log in on Klakix's own page (OAuth with PKCE). The assistant never sees your password; it gets an access token that lasts an hour and is renewed in the background. Tokens are stored only as hashes.
  • Two clear permissions

    mcp:read lets the assistant see your balance, the models and your library. mcp:generate lets it start generations that spend credits. Both are listed on the consent screen before you approve.
  • Disconnect any time

    The MCP page in the studio lists every connected app and API key, with recent activity. Remove one and it stops working right away.
  • The same rules as the web

    Content checks, credit limits and consent work exactly as in the studio. Tools that use a photo of a person show a consent statement in the chat and wait for you to confirm it before the photo is used.

FAQ

Common questions

What is an MCP server?

MCP (Model Context Protocol) is an open standard Anthropic introduced in November 2024 that lets AI assistants use outside tools. An MCP server lists what it can do; an app such as Claude, ChatGPT or Cursor calls those tools when your request needs them. The Klakix MCP server gives your assistant 19 tools for video, images and audio.

Is it free?

Connecting is free, but generating isn't. Every video, image or voiceover uses credits from your Klakix account at the same price as on the web. Before your first credit purchase only voiceover works; video, image and talking-video tools become available once you buy credits.

Learn more: Pricing

Do I need a subscription?

No. Any credit pack works, starting at $5 for 75 credits, and credits never expire. Plans add monthly credits and more generations at once. Higgsfield's MCP, by comparison, needs an active paid Higgsfield subscription.

How do I add it to Claude, and which plans support it?

In Claude on the web or desktop, open Customize > Connectors, click + Add, choose Add custom connector, paste https://klakix.com/api/mcp and log in to your Klakix account. Custom connectors work on every Claude plan; the Free plan allows one custom connector, and on Team and Enterprise an Owner adds it for the organization. Once added, it also works in the Claude mobile app.

How do I add it to ChatGPT?

On ChatGPT on the web, open Plugins (chatgpt.com/plugins), click + and choose Add custom MCP server, paste the server URL and pick OAuth. Confirm the warning, create the plugin, install it and log in. Business, Enterprise and Edu workspaces have full support; on personal plans, whether creation tools run depends on OpenAI's current rules for custom servers.

Do I need an API key?

Not for Claude, ChatGPT, Cursor, VS Code and the other apps on this page: they connect by logging in to your account (OAuth). For tools without OAuth, such as n8n, Make or your own scripts, create an API key on the MCP page in the studio and send it as an Authorization: Bearer header. The key is shown only once.

How are credits charged, and can I see the cost first?

Each generation costs the same credits as in the studio. Ask the assistant what something will cost and it calls estimate_cost before generating. Failed generations are refunded automatically, and cancel_generation refunds a job that's still in the queue.

Which models can I use?

Video: Seedance 2.5, Seedance 2.0, Seedance 2.0 Mini, MiniMax H3 Max Turbo, MiniMax H3 Max, Google Gemini Omni Flash, Google Veo 3.1 Lite, Wan 3.0, Kling 4.0, and Kling 3.0. Images: Muse Image, Nano Banana 2, and GPT Image 2.5. Talking videos: MiniMax H3 Max Lip Sync, HeyGen Avatar IV, and OmniHuman 1.5. Plus voiceover, multi-voice dialogue, music and sound effects, animated explainers and video upscaling. Ask the assistant to call list_models for the live list with prices.

Learn more: AI video models

Where do the finished files go?

The assistant gets a link to each finished file and shows it in the chat. Every result is also saved in your Klakix library, the same as anything you make in the studio.

How is it different from Higgsfield MCP?

Both connect over OAuth and charge credits for every generation. Klakix doesn't need a subscription, charges MCP generations at its web prices, and adds talking videos, animated explainers and ready-made templates. Higgsfield offers more models, its own Soul characters and voice cloning.

Guides

Step by step for each app

Connect your assistant

Sign up, buy any credit pack (from $5), then add Klakix to Claude, ChatGPT or Cursor. Your connections and API keys live on the MCP page in the studio.

Open MCP settings