How to Use Kling 3.0 Standard in Claude Code and Cursor?

Kling 3.0 Standard can be run from Claude Code or Cursor after you connect the Makernice MCP server; the agent quotes the clip with estimate_cost and only calls generate once you accept, passing that quote as max_cost_usd. Standard is billed at 720p, and 5 seconds costs $0.6048 on price list v15 (November 12, 2026).

How do I add Kling 3.0 Standard to Claude Code and Cursor?

Kling 3.0 Standard needs one Makernice MCP connection per editor, which takes a couple of minutes. For Claude Code, the Claude Code connection guide gives you claude mcp add --transport http --scope user makernice and the server address; run the command, then use /mcp in Claude Code to sign in. For Cursor, the Cursor connection guide has a one-click install and the ~/.cursor/mcp.json alternative; sign in from Cursor's MCP settings afterwards.

If you're not sure the connection worked, ask the agent for get_balance. A reply with your balance and the connection's per-job limit means the tools are live.

Can I make Kling 3.0 Standard clips in Claude or ChatGPT instead?

Kling 3.0 Standard clips can be made in Claude and ChatGPT as well, and only the connection differs. Once Makernice is added, the chat app calls estimate_cost, waits for your go-ahead, sends generate with that figure as max_cost_usd and reads the result from check_job, the same three calls Claude Code and Cursor make for Kling 3.0 Standard. The estimate is not a reserved price: generate quotes again and stops if the new quote is over your limit.

• Claude: Customize → Connectors → + Add → Add custom connector, paste the MCP URL, then Connect and sign in. On Team and Enterprise plans an owner adds it in Organization settings first. Claude steps.
• ChatGPT on the web: Plugins → + → Add custom MCP server, Authentication set to OAuth and the ChatGPT-only URL, then type @ in a new chat and pick Makernice. Your plan or workspace must allow custom MCP servers, and that version has no get_topup_link tool, so top up on the site. ChatGPT steps.

How do I animate a product photo with Kling 3.0 Standard?

Kling 3.0 Standard animates a photo in image-to-video mode: the agent puts the photo link in image_url, writes the motion into the prompt and leaves aspect_ratio out, because the clip follows the photo's shape. The exchange below turns a vertical sneaker shot into a 5-second spin.

You: Take the vertical sneaker photo from our catalog bucket and make it spin slowly under a spotlight, using Kling 3.0 Standard. Price?

Cursor calls estimate_cost with the photo link and no aspect ratio, since image-to-video follows the photo:

{
  "name": "estimate_cost",
  "arguments": {
    "input": {
      "model": "kling-3.0-standard",
      "mode": "i2v",
      "prompt": "The sneaker rotates slowly on its pedestal, a soft spotlight sweeps across the laces, tiny dust particles float in the beam",
      "image_url": "https://example.com/catalog/sneaker-vertical.jpg",
      "duration": 5
    }
  }
}

Cursor: Kling 3.0 Standard, 5 seconds, billed at 720p: $0.6048 on price list v15. Nothing is charged until you say so.

You: Do it.

Cursor calls generate with the same input, the approved max_cost_usd and a request_id for this product:

{
  "name": "generate",
  "arguments": {
    "input": {
      "model": "kling-3.0-standard",
      "mode": "i2v",
      "prompt": "The sneaker rotates slowly on its pedestal, a soft spotlight sweeps across the laces, tiny dust particles float in the beam",
      "image_url": "https://example.com/catalog/sneaker-vertical.jpg",
      "duration": 5
    },
    "max_cost_usd": "0.6048",
    "request_id": "sneaker-spin-i2v-std"
  }
}

Makernice quotes the request again on arrival and creates the job only when that quote is within max_cost_usd and the connection's per-job limit; the charge is that quote. Above either line, the call is refused and nothing is created. Should Cursor retry the identical call with the same request_id, it receives the job it already started.

Cursor then polls check_job with the job_id, passing wait_s (55 maximum) to hold the call until the clip is done. The MP4 sits behind outputs[0].url, a signed link valid for 24 hours that a new check_job call refreshes; console_url opens it in the Makernice console.

Framing, length and quality choices on Kling 3.0 Standard

Aspect ratio only applies to text-to-video, which offers 9:16, 16:9, 1:1. In image-to-video and first and last frame the clip takes the shape of your image, and adding aspect_ratio gets the request refused. To get vertical video from a photo, start from a vertical photo.

Kling 3.0 Standard has one tier, billed at 720p for $0.1209 per second on price list v15 (November 12, 2026). Leave resolution out of the request. For a 1080p version of the same idea, change the model id to Kling 3.0 Pro; the Pro vs Standard comparison walks through when each makes sense. A 3-second vertical text-to-video request on Standard looks like this:

{
  "name": "estimate_cost",
  "arguments": {
    "input": {
      "model": "kling-3.0-standard",
      "mode": "t2v",
      "prompt": "A white sneaker drops onto a concrete floor in slow motion and bounces once, vertical framing, hard side light",
      "duration": 3,
      "aspect_ratio": "9:16"
    }
  }
}

Durations run from 3 to 15 seconds (5 if unset). Prompts can be at most 2500 characters. There are three modes, text-to-video, image-to-video and first and last frame; reference-to-video isn't offered. Sound is always part of the clip. The Kling 3.0 Standard model page has the complete specification.

Input photos are sent as public https links to JPEG or PNG files of no more than 10 MB (Kling does not take WebP). Kling 3.0 Standard needs the shorter side to be at least 300 px and the width-to-height ratio to sit between 0.4 and 2.5, so a very tall banner image may need cropping. Local files aren't readable over MCP.

Kling 3.0 Standard refusal messages explained

A refusal is a plain-text reply; no job was created and you were not charged. Column one names the refusal as Makernice's server code does; column two shows how the message begins.

If a job fails once it has started, for instance after moderation rejects the input photo, check_job reports status failed and billing released, and the held amount returns. More on Kling failures: why Kling generations fail.