Kling AI 3.0: what a clip costs and how to use it

# Kling prompt card · AIGE

How to write a Kling prompt that works on the first run, and which Kling to pick in AIGE (aige.ws).

Updated 2026-09-27 · By AIGE · Prices in AIGE credits (cr). The studio shows the exact cost before every run.

---

## Which Kling to pick

| Kling in AIGE | Pick it for | What you get in AIGE | 5 s costs |
|---|---|---|---|
| **Kling V3** (Kling VIDEO 3.0) | The default. Characters, products, anything with sound | 3–15 s, sound on or off, start frame and end frame | Standard: 63 with sound, 42 silent · Pro: 84 with sound, 56 silent |
| **Kling 3 Turbo** | Fast turnaround (how fal describes Turbo Standard) | 3–15 s, Pro renders 1080p, Standard 720p. Start frame only, no sound switch | Pro 70 · Standard 56 |
| **Kling o3** (Kling VIDEO 3.0 Omni) | Kling's Omni model on a prompt, start frame or start + end frame | 3–15 s, sound switch (off by default in the studio) | 70, whether Pro or Standard, with or without sound |
| **4K** (switch on V3 or o3) | The one hero shot that has to hold up on a big screen. Kling renders it natively in 4K, not upscaled | Same controls as V3 / o3 | 210, with or without sound |
| **Kling Motion Control** | Putting the movement from a reference clip onto your character picture | Length follows your clip: up to 10 s (Character: Image) or 30 s (Character: Video). Pro or Standard, keep the clip's sound | Pro 84 · Standard 63 per 5-second step of clip |

**Pro or Standard.** Kling describes Standard as faster and cheaper to run, and Professional as richer in detail, with more refined composition and color.
- In the web studio, Kling V3 and o3 have no Pro/Standard switch of their own. They follow the Quality you last set on a video model that has one (Kling 3 Turbo, Veo, Seedance), and that Quality is **Standard by default**.
- The cost next to Generate shows which tier you're on. For a 5 s V3 clip with sound, 63 means Standard and 84 means Pro.
- From Claude or ChatGPT you pick it by name: `kling-v3` is Pro and `kling-v3-fast` is Standard.

**Worth knowing before you choose**
- Kling 3 Turbo costs more than Kling V3 with sound off (5 s: Turbo Standard 56 vs V3 Standard silent 42).
- On o3 and on 4K, turning sound off doesn't lower the price. Only Kling V3 has a cheaper silent rate.
- On the AIGE Canvas timeline, Kling clips run at Standard (Kling V3 and o3 with sound on). Kling 3 Turbo isn't offered for "Between two shots" because it takes no end frame.
- Plan limits: under AIGE's built-in plan policy the free plan can't run Kling, the entry paid plan runs 5-second Kling V3, Turbo and o3 clips only, and 4K and Motion Control need a higher plan.

---

## The prompt formula

Kling's own text-to-video formula:

**Subject** (with description) **+ Subject movement + Scene** (with description) **+ optional: Camera language + Lighting + Atmosphere**

| Slot | What goes in it | Example |
|---|---|---|
| Subject | Who or what the shot is about. Add looks in short sentences: hair, clothes, posture | *A man in his 60s with a grey beard and a navy fisherman's sweater* |
| Subject movement | What the subject does. Keep it plain and short enough to happen within the clip | *slowly coils a rope and looks out to sea* |
| Scene | Where it happens: foreground, background, indoor or outdoor | *on the deck of a small fishing boat, harbour lights behind him* |
| Camera language | Shot size, angle, lens and camera move (see the next section) | *medium shot, slow push in, shallow depth of field* |
| Lighting | The light source and its quality | *blue hour, warm lamp light on his face* |
| Atmosphere | The mood you want | *quiet, reflective* |

Subject, movement and scene are the core. Camera, lighting and atmosphere are optional.

With sound on, Kling's audio guide extends the formula: **Scene + Element (subject) + Movement + Audio** (dialogue, singing, sound effects or music) **+ Other** (style, emotion, camera).

**Rules from Kling's guides**
- Use simple words and simple sentences.
- Keep what's in the frame simple enough to finish within the clip.
- Describe each subject and the scene in a few short sentences, not one long one.
- Kling 3.0 renders up to 15 s in one generation. Kling's own 15-second long-take example paces the action with time cues, e.g. "At the 4th second…", "In the final 3 seconds…".
- Kling's Multi-Shot storyboard isn't available in AIGE. For several shots, make one clip per shot and cut them together on the Canvas timeline.
- Prompt length: Kling's endpoint takes up to 2,500 characters. AIGE's Claude/ChatGPT connector takes up to 2,000.

---

## Camera language Kling understands

Every phrase below comes from Kling's own guides and example prompts. Write the camera move into the prompt. Kling's camera-control presets aren't in AIGE.

| Kind | Phrases Kling uses |
|---|---|
| Shot size | close-up, macro shot, medium shot, medium-long shot, wide shot, mid-close-up |
| Angle | low-angle, high-angle, eye-level, profile shot, frontal, top-down / bird's-eye view, aerial shot, first-person POV |
| Camera move | tracking shot (tracking behind / tracking backward in front of), the camera follows, zooms in, pans slowly in, pushes in over the shoulder, pulls back, orbits / circles, 360-degree pan, rises, gentle downward tilt, slowly advances forward |
| Holding still | the camera lingers on, remains fixed on, freezes when the subject pauses |
| Style | handheld / cinematic handheld, one continuous long take without cuts, slow motion, motion blur |
| Lens and look | ultra-wide angle, telephoto, bokeh / blurred background, depth of field, tilt-shift, symmetrical composition, cinematic color grading |

---

## Motion and physics tips

- **Write the action, not the end state.** Kling's example: "wear sunglasses" is vague, but "puts on sunglasses with her hand" works. Still photos tend to come out static when the prompt doesn't describe movement.
- **Keep motion physically plausible**, and pick movement that could really happen in the scene.
- **Kling says some motion is still hard:** a bouncing ball, or the arc of an object thrown high. Plan around it, or keep those beats off-screen.
- **Don't rely on counts.** Kling says video models aren't sensitive to numbers, so "10 puppies on the beach" won't reliably give ten.
- **Several subjects:** give each one its own movement and list them in order.
- **Split screen:** Kling suggests phrasing like "4 camera angles, representing spring, summer, autumn, and winter".

---

## Sound

Sound is on by default on Kling V3 in AIGE. **Switching it off lowers the price on V3:** a 5 s clip goes from 84 to 56 cr at Pro and from 63 to 42 cr at Standard. On o3 and 4K the price is the same either way. Kling 3 Turbo has no sound switch in AIGE.

How Kling's guides tell you to prompt sound:
- **Dialogue.** Give a character label, emotion and tone, then the line in quotation marks: `[Barista, warm, relaxed voice] says: "your usual? coming right up."`
  - Put the action first and the line after it, so the model knows who is speaking.
- **Two or more speakers.** Kling's guide gives four rules:
  1. Give each character a unique label and keep it the same every time. Don't switch to "he" or "she".
  2. Tie each line to that character's action.
  3. Give each voice its own tone.
  4. Mark the handover with a linking word like "Immediately," so one voice doesn't read every line.
- **Sound effects.** Give the source, the action and the sound: *a glass slips and shatters on the tile floor, a sharp crash*.
- **Ambience.** Give the scene, the sound elements and the space: *a train station at night, distant announcements and footsteps, echoing through the big hall*.
- **Music.** Give the instrument, the genre and the emotion: *acoustic guitar, folk, hopeful*.
- **Use quotation marks** around anything that should be heard.
- **Language and spelling.**
  - fal's page for the Kling V3 endpoint AIGE calls lists Chinese and English voice output, and other languages are translated to English.
  - For English speech, write words in lowercase and use uppercase only for acronyms and proper nouns.
  - Kling's 3.0 guide also shows accent tags, e.g. "speaks in English with an Indian accent".
- **Give dialogue room.** Kling's audio guide recommends the longer 10-second setting for singing or dialogue.

---

## Starting from an image

- **Formula for image-to-video:** **Subject + Movement, Background + Movement**. The image already sets the scene, so spend the words on what moves.
- **Stay close to the picture.** Kling warns that a prompt that strays far from the image can cause a camera cut or transition.
- **Use a sharp, high-resolution image.** Kling says image-to-video quality depends heavily on the input resolution.
- **Image limits** for the Kling endpoints AIGE calls: at least 300 px on each side, up to 50 MB, and a shape between 1:2.5 and 2.5:1.
- **Aspect ratio comes from the image.** With a start frame, the clip takes the image's shape and AIGE's ratio setting doesn't apply. Without an image, Kling V3, Turbo, o3 and 4K take 16:9, 9:16 or 1:1.
- **Text on products.** Kling 3.0 is built to keep text from the uploaded image, such as signs, captions and logos, legible and in place.
- **End frame (Kling V3, o3 and 4K; not Turbo).** Choose two similar images with the same subject. Kling says big differences between them can trigger a shot switch.
- **Your AIGE character** gives Kling the start frame. Kling's own "element binding" and multi-image references aren't available in AIGE.

---

## Motion Control

Motion Control copies the movement in a reference clip onto the character in your picture.

**Your character picture**
- Show clear body proportions with nothing covering the character, and make the character more than 5% of the image.
- Match the framing to the clip: full body with full body, half body with half body.
- For big moves, leave room in the picture for the character to move.
- Limits: 340–3,850 px, up to 10 MB, shape between 1:2.5 and 2.5:1.

**Your reference clip**
- One person, whole body or upper body with the head visible and not blocked. Kling recommends real human movement.
- With several people in the clip, Kling uses whoever takes up the most of the frame.
- One continuous shot: no cuts, no shot changes and no camera moves, or the result may be cut short.
- Steady, moderate movement works best. With very fast or complex motion the output can come out shorter than your clip.
- Limits: 3–30 s long, 340–3,850 px, up to 100 MB.

**Character: Image or Video**
- **Image** (the studio default) keeps the orientation of your picture and allows clips up to 10 s. You can steer the camera in the prompt: Kling lists zoom in, zoom out, camera up, camera down and fixed position.
- **Video** follows the clip's orientation and camera. The endpoint docs call it better for complex motion, and it allows clips up to 30 s.

**Other settings**
- **Prompt:** use it for the background and extra details, e.g. *a small terrier runs in and circles her feet*.
- **Original sound:** on by default. It keeps the audio of your reference clip.
- **Cost:** AIGE measures your clip and bills it in 5-second steps, rounded up (Pro 84, Standard 63 per step). Trim a 5.5 s clip to 5 s and it bills as 5 s instead of 10 s.
- **Too-long clips:** a clip over the limit for your Character setting is stopped before anything is charged.

---

## Common problems and fixes

| Problem | What Kling's docs say | Fix |
|---|---|---|
| The photo barely moves, or the camera just pans over it | Photos tend to come out static when no action is described | Write subject + action ("she lifts the cup and blows on it") |
| An unwanted cut in image-to-video | A prompt far from the image can cause a cut or transition | Describe what's in the picture; keep new elements small |
| A cut between the start and end frame | Big differences between the two frames can trigger a shot switch | Use two similar images with the same subject |
| Wrong number of things | Models aren't sensitive to numbers | Avoid exact counts; frame fewer subjects |
| A ball or throw looks wrong | Complex physical motion is still hard | Keep the action simple and physically plausible |
| The wrong person speaks, or one voice reads all the lines | Speakers need unique labels, actions and linking words | Use `[Name, tone]: "line"`, then "Immediately," before the next speaker |
| Speech sounds off | The endpoint voices Chinese and English, and wants lowercase English | Write English lines in lowercase, with acronyms in capitals |
| A soft image-to-video result | Output quality depends on input resolution | Start from a sharper, larger image |
| A Motion Control result shorter than the clip | Fast or complex motion is cut to the usable part | Use steady movement, one shot, no camera moves |
| Your ratio setting was ignored | With a start frame the ratio comes from the image | Crop the image to the shape you want before you start |

---

## 8 example prompts

These examples follow Kling's structure, with each part of the formula in its own sentence. Costs are for the settings listed.

**1 · Product turntable** (start from your product photo)
```
The black ceramic mug with the gold "NORTH" logo stands still on a round marble pedestal. The camera orbits the mug in slow motion, one smooth full circle. Seamless warm-grey studio backdrop. Close-up, shallow depth of field, soft key light from the left with a thin rim light, calm premium mood.
```
Kling V3 · start frame · 5 s · sound off · Standard. **42 cr** (Pro: 56)

**2 · Character walking shot**
```
A young woman with a short silver bob, oversized camel wool coat and white sneakers walks toward the camera along a rain-wet city sidewalk at dusk, hands in her pockets, a small smile. Shop windows glow behind her and passing cars blur. Tracking shot moving backward in front of her, medium shot, neon reflections on the wet pavement, cinematic color grading. Audio: her footsteps on wet pavement, distant traffic noise, light rain.
```
Kling V3 · text-to-video · 9:16 · 8 s · sound on · Standard. **101 cr**

**3 · Food close-up**
```
A cast-iron skillet on a gas flame, two eggs frying beside thick-cut bacon, the egg whites bubbling at the edges as the bacon curls. Rustic wooden counter behind, soft morning window light, a thin wisp of steam. Macro shot, shallow depth of field, slow push in. Audio: loud sizzling, oil bubbling, quiet kitchen room tone.
```
Kling V3 · text-to-video · 16:9 · 5 s · sound on · Standard. **63 cr**

**4 · Fashion / UGC try-on**
```
A bright bedroom with a full-length mirror and a clothes rack. [Maya, a young woman with curly dark hair in a cropped denim jacket] turns once to show the fit, then steps closer to the camera. [Maya, cheerful, casual voice] says: "okay, this jacket goes with literally everything. light enough for spring, and the pockets are huge." Selfie-style vlog close-up, natural daylight. Background: soft upbeat music.
```
Kling V3 · text-to-video (or start from your saved character) · 9:16 · 10 s · sound on · Standard. **126 cr** (Pro: 168)

**5 · Cinematic establishing shot**
```
Aerial shot slowly advancing over a fjord at sunrise, a small red wooden cabin on the rocky shore, low mist drifting over dark water, snow-capped mountains on both sides. Ultra-wide angle, soft golden light breaking through clouds, calm and epic atmosphere, cinematic color grading. Audio: wind over open water, distant seabirds.
```
Kling V3 · text-to-video · 16:9 · 8 s · sound on · 4K on. **336 cr**. Draft it first at Standard (5 s with sound: 63 cr), then run the keeper in 4K.

**6 · Portrait comes to life** (start from a portrait)
```
The woman slowly turns her head toward the camera and breaks into a warm smile, a loose strand of hair lifting in the breeze. Behind her, leaves sway and sunlight flickers through them. The camera lingers on her face. Audio: soft breeze, leaves rustling, distant birdsong.
```
Kling o3 · start frame · 5 s · sound on (same price as off) · **70 cr**

**7 · Action / sport**
```
A trail runner in a neon-orange windbreaker sprints down a rocky mountain ridge, arms pumping, loose gravel kicking up behind each stride. Low-angle side tracking shot keeping pace with her, motion blur on the foreground rocks, harsh late-afternoon sun, intense and determined mood.
```
Kling 3 Turbo · text-to-video · 16:9 · 5 s · Pro (1080p) · **70 cr**. With sound on Kling V3 Standard, add: *Audio: heavy breathing, gravel crunching underfoot, strong wind.* (63 cr)

**8 · Social ad hook**
```
Iced coffee is poured over a tall glass of ice on a sunny café table, dark coffee swirling into the milk, condensation running down the glass. [Off-screen voice, crisp, upbeat female voice]: "stop scrolling. your afternoon just got better." Close-up, bright natural light, vivid colors. Background: ice clinking, a short upbeat beat.
```
Kling V3 · text-to-video · 9:16 · 5 s · sound on · Standard. **63 cr**

---

## Settings and prices in AIGE

| Kling in AIGE | Lengths | Start frame | End frame | Sound | Text-to-video ratios |
|---|---|---|---|---|---|
| V3 (Pro or Standard) | every second, 3–15 s | yes | yes | switch, on by default | 16:9, 9:16, 1:1 |
| 3 Turbo (Pro 1080p / Standard 720p) | every second, 3–15 s | yes | no | no switch | 16:9, 9:16, 1:1 |
| o3 (Pro or Standard) | every second, 3–15 s | yes | yes | switch, off by default in the studio | 16:9, 9:16, 1:1 |
| V3 4K / o3 4K | every second, 3–15 s | yes | yes | switch (same price) | 16:9, 9:16, 1:1 |
| Motion Control (Pro or Standard) | follows your clip, 3–10 s (Image) or 3–30 s (Video) | your character picture | no | keep clip's sound, on by default | not applicable |

| Price per clip (cr) | 5 s | 10 s | 15 s |
|---|---|---|---|
| V3 Pro, with sound | 84 | 168 | 252 |
| V3 Pro, silent | 56 | 112 | 168 |
| V3 Standard, with sound | 63 | 126 | 189 |
| V3 Standard, silent | 42 | 84 | 126 |
| 3 Turbo Pro (1080p) | 70 | 140 | 210 |
| 3 Turbo Standard (720p) | 56 | 112 | 168 |
| o3 (Pro or Standard, sound on or off) | 70 | 140 | 210 |
| V3 4K or o3 4K | 210 | 420 | 630 |
| Motion Control Pro (billed on your clip's length) | 84 | 168 | 252 |
| Motion Control Standard | 63 | 126 | 189 |

Any whole second from 3 to 15 s can be set; the studio shows the exact cost of your length before you run. Motion Control bills in 5-second steps of your reference clip.

---

## Use it from Claude or ChatGPT

AIGE has a connector (MCP server) at **`https://aige.ws/mcp`**.
- **Claude:** go to Settings, then Connectors, and choose Add custom connector with that URL.
- **ChatGPT:** turn on developer mode under Connectors, create a connector with that URL and choose OAuth.
- **Sign in** on the AIGE screen with an API key or your AIGE email and password. Setup for every client is on aige.ws/mcp-info.

The tool is **`antix_generate_video`**. The fields that matter for Kling:

| Field | Value for Kling |
|---|---|
| `model` | `kling-v3` (Pro) or `kling-v3-fast` (Standard) |
| `prompt` | your prompt, up to 2,000 characters |
| `duration` | whole seconds, 3–15 |
| `audio` | `true` (default) or `false` (cheaper) |
| `aspectRatio` | for text-to-video: 16:9, 9:16 or 1:1 (other portrait ratios become 9:16, landscape 16:9) |
| `sourceImage` | start frame (HTTPS URL or base64 data URL) |
| `endImage` | end frame; needs a start frame (`sourceImage` or `characterId`) |
| `characterId` | a saved AIGE character, which supplies the start frame |
| `count` | 1–4 separate clips, each charged |

Only Kling V3 Pro and Standard are available through the connector. Kling 3 Turbo, o3, 4K and Motion Control are in the web studio. The call returns a task id, and the assistant checks it with `antix_check_task`. If you're unsure which model to use, `antix_pick_model` is free.

**A realistic ask**

> Use AIGE to make a 5-second Kling V3 clip with sound from this photo: https://example.com/mug.jpg. Prompt: "The black ceramic mug stands still on a marble pedestal as the camera orbits it in slow motion. Soft key light, shallow depth of field. Audio: quiet café ambience, a spoon clinking against a cup."

The assistant should send:

```json
{
  "model": "kling-v3",
  "prompt": "The black ceramic mug stands still on a marble pedestal as the camera orbits it in slow motion. Soft key light, shallow depth of field. Audio: quiet café ambience, a spoon clinking against a cup.",
  "duration": 5,
  "audio": true,
  "sourceImage": "https://example.com/mug.jpg"
}
```

Cost: **84 cr** (Kling V3 Pro, with sound, 5 s). With `"model": "kling-v3-fast"` it runs at Standard, and with `"audio": false` it's cheaper still.

---

## Sources

1. Kling AI, Kling VIDEO 3.0 Model User Guide: https://kling.ai/quickstart/klingai-video-3-model-user-guide
2. Kling AI, Kling VIDEO 3.0 Omni Model User Guide: https://kling.ai/quickstart/klingai-video-3-omni-model-user-guide
3. Kling AI, Text-to-Video prompt guide: https://kling.ai/quickstart/text-to-video-prompt-guide
4. Kling AI, Image-to-Video guide: https://kling.ai/quickstart/image-to-video-guide
5. Kling AI, Start and End Frames: https://kling.ai/quickstart/ai-video-start-end-frames
6. Kling AI, Standard Mode & Professional Mode: https://kling.ai/quickstart/how-to-choose-video-mode
7. Kling AI, KLING VIDEO 2.6 native audio guide (the audio model 3.0 builds on): https://kling.ai/quickstart/klingai-video-26-audio-user-guide
8. Kling AI, Motion Control user guide: https://kling.ai/quickstart/motion-control-user-guide
9. Kling AI, Camera Movement: https://kling.ai/quickstart/ai-camera-control-guide
10. Kling AI, native 4K announcement: https://kling.ai/blog/kling-ai-introduces-native-4k-video-model
11. fal, Kling V3 Pro image-to-video API (AIGE's V3 endpoint): https://fal.ai/models/fal-ai/kling-video/v3/pro/image-to-video/api
12. fal, Kling V3 Turbo Pro image-to-video API: https://fal.ai/models/fal-ai/kling-video/v3/turbo/pro/image-to-video/api
13. fal, Kling V3 Turbo Standard image-to-video: https://fal.ai/models/fal-ai/kling-video/v3/turbo/standard/image-to-video
14. fal, Kling O3 Pro image-to-video API: https://fal.ai/models/fal-ai/kling-video/o3/pro/image-to-video/api
15. fal, Kling V3 4K image-to-video API: https://fal.ai/models/fal-ai/kling-video/v3/4k/image-to-video/api
16. fal, Kling V3 Pro Motion Control API: https://fal.ai/models/fal-ai/kling-video/v3/pro/motion-control/api
17. fal, Kling 3.0 and O3 overview: https://fal.ai/kling-3
18. AIGE, connector setup: https://aige.ws/mcp-info