Skip to main content
Unifically LogoUnificAlly
Model logo

Kling 2.5 Turbo API

  • Text to Video
  • Image to Video
Click or drag & dropPNG, JPG, WEBP, GIF · Max 100MB
Generate Audio
Generate audio for the output video (default: false)
Output

Your generated video will appear here

Kling 2.5 Turbo

What is Kling 2.5 Turbo?

Kling 2.5 Turbo is the speed-and-cost variant of Kling's 2.x line. The parameter surface is close to Kling 2.6: text-to-video or image-to-video on the same model ID, 5 or 10 second clips, Standard (720p) or Pro (1080p) output, three aspect ratios in the text path, and a sound effects track in the same call. The trade is a faster queue and a lower per-clip cost in exchange for slightly less output quality. Use it for prompt sweeps, draft client reviews, and ops tests where you want to run many variations cheaply before the final render. For delivery, jump to Kling 2.6 or 3.0.

Key features of Kling 2.5 Turbo

Five features define what production teams use Turbo for.

Text-to-video and image-to-video on one model

Pass a prompt on its own for fresh generation, or add a start frame to animate a supplied still. Same model ID, no branching call sites in your client code.

Standard or Pro output modes

Standard renders at 720p for cheap iteration. Pro renders at 1080p when the cut is heading to a hero unit or paid placement.

Three aspect ratios in the text path

16:9, 9:16, or 1:1 for prompt-only generation. The aspect_ratio field is ignored when a start frame is set; output inherits the image native ratio.

Sound effects in the same call

The sound_effects object accepts an ambient sound prompt, a music prompt, or asmr_mode for ASMR-style audio. Skip the field for a silent clip.

Best for

Prompt sweeps

Generate ten variations of the same brief cheaply. Useful for tuning prompt wording before the final delivery render.

Storyboard motion previews

Rough MP4s from key frames before committing the shot to a heavier model.

Operations and load tests

Lighter model for queue and latency testing in CI/CD pipelines.

Draft client reviews

Fast MP4s for direction calls. Final grade later on Kling 2.6 or 3.0.

Internal-facing content

Internal demos and stand-up summaries where final quality is not the bar.

Use cases

Run a prompt sweep across ten variations of a single brief by holding the start frame constant and changing prompt wording. Pick the strongest two and rerun on Kling 2.6 Pro for delivery. Load-test a video pipeline before launch by replaying a representative job mix on Turbo Standard, where each call is cheap enough to do hundreds of times. Build a quick draft cut for a client direction call by passing the boards as start frames at 720p, then escalating only the approved shot to a heavier model.

API examples

Call Kling 2.5 Turbo from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/video/kling/kling-2.5.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "kuaishou/kling-2.5-turbo-video",
    "input": {
      "prompt": "Cinematic scene with dramatic lighting",
      "duration": 10,
      "mode": "pro"
    }
  }'

Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.

FAQs

People also ask

Kling 2.5 Turbo is the speed-and-cost variant of the 2.x line. It runs text-to-video and image-to-video on the same model ID and outputs 5 or 10 seconds at Standard (720p) or Pro (1080p). A sound effects track is available in the same call.

The parameter surface is almost the same. Both support text-to-video, image-to-video, 5 or 10 second clips, Standard or Pro, and a sound effects object. The 2.6 endpoint adds native audio with voice references; Turbo keeps the audio path on the simpler sound_effects object.

Use Turbo for short 5 to 10 second loops and prompt iteration. Move to Kling 3.0 when you need multi-shot storytelling, clips up to 15 seconds, 4K output, or native AI audio with voice references.

No. The schema does not expose a negative prompt or a seed. Steer the output with prompt wording, the start frame upload, mode, and aspect ratio.

Through the sound_effects object. Pass a sound prompt, a music prompt, or enable asmr_mode for an ASMR-style track. Omit the field entirely for a silent output.