Wireflow is now a Claude connector.

Set it up→

Kling in Claude: How to Run Kling 4.0 From a Chat

Andrew Adams

Andrew Adams

·10 min read
Kling in Claude: How to Run Kling 4.0 From a Chat
In this article▾
  1. What "Kling in Claude" means
  2. How Claude calls Kling over MCP
  3. The run: one photo, Kling 4.0, 5 seconds
  4. What came back
  5. Set it up and run it from Claude
  6. When to move from Claude to the Kling API
  7. Costs and limits to plan for
  8. FAQ
  9. Conclusion

Kling in Claude means asking Claude for a video and getting a real Kling 4.0 clip back, made from your own photo. This guide covers the connector setup, one image-to-video run and what it cost. Claude has no video model of its own, so it calls one through MCP. We used Wireflow, a node-based AI canvas whose connector lets Claude run a saved Kling 4.0 flow and return the MP4. On 28 October 2026 our 5-second clip from one product photo took just under 4 minutes and charged 91 credits.

What "Kling in Claude" means

Claude reads, writes and calls tools, but it does not render video. Kling is Kuaishou's video model family, and Kling 4.0 is its newest release: 3 to 30 second clips at 720p or 1080p, from text, a start frame, an end frame or up to 10 reference images. To get a Kling clip from a chat, Claude needs an MCP server that exposes Kling as a tool.

The pages that rank for this phrase describe three kinds of setup:

SetupWho holds the Kling accessWhat Claude can do
Self-hosted Kling MCP serverYou, with your own Kling API keyStart a Kling job and poll it
Hosted Kling-only connectorThe host, paid with its creditsKling text or image to video
Workflow connectorYour account, signed in with OAuthRun a saved flow with Kling 4.0 as one node, and call the same flow over REST

Most of those guides stop at installation and cover older Kling versions. For a wider look at connector options, see our ranked list of video MCP tools.

Try it yourself

Build this in Wireflow

Make the same kind of asset on a canvas you control, with a free account and no credit card.

How Claude calls Kling over MCP

The Model Context Protocol (MCP) lets an app publish tools that Claude calls by name. A video takes minutes, so the tool starts a job, returns an id, and Claude checks back until a file URL appears. Our test of Claude video generation used the same loop with other models.

The MCP server we connected has these tools for a Kling clip:

  • list_models and get_credit_balance show prices and the balance before anything is spent.
  • list_workflows and get_workflow find a saved flow and the inputs it takes.
  • create_upload_link sends you a link to upload a photo (opt-in permission).
  • run_workflow starts a run. It is the only tool that spends credits.
  • get_execution returns the status and, once done, the output URLs.

The run: one photo, Kling 4.0, 5 seconds

Most people want Kling for a short video from a still, such as a moving product shot. Our flow is a plain image to video AI setup in three parts:

  1. Start Frame (photo). An image input. Ours is a product-style photo of iced coffee with milk swirling in, made for this test.
  2. Motion Prompt. A text input that says what moves and how the camera moves.
  3. Clip (Kling 4.0). The Kling 4.0 node at 5 seconds and 720p. With only a start frame wired, the clip takes the photo's shape.

The start frame we gave Kling 4.0: a glass of iced coffee with milk swirling in, on a pale stone counter by a window

The motion prompt, word for word:

The cold milk keeps swirling slowly down through the iced coffee and blends in, the ice cubes shift and settle, and a bead of condensation runs down the outside of the glass. Slow, gentle push-in toward the glass. Soft daylight from the window, calm and still. Sound: ice cubes clinking softly, quiet room tone. No text, no logos, no people.

A start frame already fixes the subject, light and framing, so the prompt only needs one physical motion and one camera move. Our Kling prompt guide has more patterns that hold up.

The 5-second Kling 4.0 clip from our run: milk swirls through a glass of iced coffee as the camera pushes in

The loop is the unedited output without sound; click it for the MP4 with audio. Figures come from the run record, and the pricing page lists the plans behind the credits.

StepSettingsFinished afterCredits
Start Frame and Motion PromptInputsunder 1 s0
Clip (Kling 4.0)5 s, 720p, start frame3 min 54 s91
Totalabout 3 min 57 s91

What came back

Frames at 0.5, 2.5 and 4.5 seconds from the Kling 4.0 clip

The file is 1280x716 at 24 frames a second, 5.04 seconds long, H.264 with a stereo AAC track. The odd height comes from our 1376x768 photo, whose shape Kling kept. For other ways to put a still into motion, see how to animate still images with AI.

  • The motion matched the prompt. The milk swirls, the ice shifts, and by 4.5 seconds the drink has blended to an even light brown while the camera pushes in.
  • It kept what was in the photo. A thin milk stream in our start frame kept pouring for about 3 seconds.
  • One visible flaw. The "bead of condensation" became a white, milky drip running down the outside of the glass from about 3 seconds on. A product shot would need a rerun.
  • The sound stops early. Quiet room tone, a burst of sharp sounds between 2 and 3.3 seconds, then near silence from about 3.8 seconds.

That drip is why a person should watch the first take: Claude can rerun without the condensation line, or try a second model on the same photo, the idea behind comparing video models side by side.

Set it up and run it from Claude

The connector signs in with OAuth, so there is no key to paste. It works in claude.ai, Claude Desktop and Claude Code, the terminal route on our AI API with Claude Code integration page.

  1. Add the connector. In claude.ai, open Customize, then Connectors, click the plus button and choose Add custom connector. Paste https://www.wireflow.ai/api/mcp and leave the advanced OAuth fields empty. In Claude Code, run claude mcp add --transport http wireflow https://www.wireflow.ai/api/mcp.
  2. Sign in and approve. Reading and running are allowed by default. Tick "Create, edit and delete workflows" only if Claude should change flows or send upload links.
  3. Copy the flow. Open the workflow linked at the end of this guide and duplicate it into your account.
  4. Ask in plain words. For example: "Run my Kling 4.0 Clip From Claude flow with this photo as the start frame. Motion: steam rises from the mug, slow push-in. Tell me the credit cost and wait for my yes." Give it a public image URL, or ask for an upload link.
  5. Let Claude poll. It quotes the price, calls run_workflow after your yes, and checks get_execution until it can reply with the MP4 link.

Claude can also write the motion prompt from the photo; read it before you say yes. To copy motion from a reference video instead, use Kling Motion Control.

When to move from Claude to the Kling API

Claude is the right front end while a person is choosing the photo, the motion and whether to rerun. It is the wrong place for work nobody watches, because the chat stays busy while a render runs. Our n8n vs MCP comparison covers that trade-off.

The flow does not change when your code takes over. Your app sends an API key as a Bearer token, POST /api/v1/workflows/{id}/run returns an execution id, and GET /api/v1/workflows/executions/{executionId}/poll returns the status and output URLs. The poll reports creditsUsed (91 for ours), so you can log each clip's cost. Request patterns and per-node cost reporting are on the Kling API page.

Claude through MCPREST API
Who starts a runYou, in a chatYour app, a schedule or a queue
Sign-inOAuth, no key to pasteAPI key as a Bearer token
Cost checkClaude reads price and balance firstYour code reads creditsUsed
Best forOne-off clips and prompt testsProduct photo batches and app features

API and MCP access are on every plan, and runs use the same credits either way. For request bodies and error handling, follow our step-by-step Kling API guide.

Costs and limits to plan for

Figures are from the live models list and pricing page on 28 October 2026. Kling's own plans are in our Kling AI pricing breakdown.

  • 19 credits a second at the 5-second rate. Our clip's free quote and its charge were both 91 credits.
  • Plans. Starter is $36 a month for 2,660 credits, about 29 of these clips. Professional is $59 a month for 4,370 credits, about 48.
  • Resolution. 1080p costs more: in our 26 October test, 5 seconds at 1080p quoted 135 credits.
  • The photo sets the shape. Crop to 16:9 or 9:16 before you run.
  • Two bills. Claude's tool calls count toward your Claude plan. The generation is charged to the connector's account.

Try it yourself: Build this workflow in Wireflow. Swap in your own photo, edit the Motion Prompt and run Clip (Kling 4.0) for a 5-second, 720p clip. From Claude, one run_workflow call does the same.

FAQ

Can Claude generate videos with Kling?

Not by itself. Claude has no video model, but through an MCP connector it can run Kling 4.0 and return the MP4 link.

Which Kling versions can Claude use this way?

Any version the flow's video node is set to. Kling 4.0 is the newest, and Kling O3 is a cheaper node at 16 credits a second, so Claude can run both on one photo to compare.

How much does one Kling 4.0 clip cost from Claude?

Our 5-second 720p clip quoted and charged 91 credits. Claude's own usage counts toward your Claude plan.

Can I use my own photo as the first frame?

Yes. Give Claude a public image URL or upload the photo through a link it sends you. The clip opens on your photo and keeps its aspect ratio.

Does the clip come with sound?

Yes, a stereo track in the same file. Ours went near silent after about 3.8 seconds, so listen before you publish.

Do I need a Kling API key?

No. The connector signs in with OAuth, and Kling runs on the account's credits. You need an API key only when your own code calls the flow, for example through a video generation API integration.

Conclusion

Kling in Claude works today through a connector: you hand Claude a photo, it quotes the price, runs Kling 4.0 after your yes and returns the file. Our 5-second test took just under 4 minutes and 91 credits, with one milky drip worth a rerun. Wireflow keeps that flow the same whether Claude or your code starts it, so the node-based video generation graph you test in a chat is the one your app calls later.

Written by
Andrew Adams

Andrew Adams

Co-Founder & Operations at Wireflow

Runs client operations and content strategy at Wireflow. Works directly with creative teams and agencies to build production AI workflows.

  • Content Strategy
  • Client Operations
Try Wireflow

Make AI videos and ads.

Start free, no card needed.

Start free