whatmodel.sh / video / Kling 3.0
Kling 3.0
↗ share
+ contribute
Kling 3.0 is a closed-weights video model from Kuaishou (Kling AI), released February 2026. Licence: proprietary.
$ how it compares
The video models ranked either side of Kling 3.0, out of 139 with a score, then more from Kuaishou (Kling AI).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 20 |  Sora 2 Pro | 68±2 | comparison only | ||
| 21 |  Veo 3.1 Fast | 68±1 | api $0.10/sec | ||
| 22 | Kling 3.0 | 68±4 | api $0.084/sec | ||
| 23 |  Wan 2.7 | 68±2 | api $0.10/sec | ||
| 24 |  Vidu Q3 Pro | 65±2 | api $0.10/sec | ||
| more from Kuaishou (Kling AI) | |||||
| 31 | Kling O3 Pro | 64±4 | api $0.11/sec | ||
| 34 | Kling 2.6 Pro | 59±2 | api $0.070/sec | ||
| 44 | Kling O1 Pro | 56±2 | api $0.11/sec | ||
$ what it's good at
image to video▁▂·▂▅█▅▄▅▆▃▄ #10 of 50 * · 38% of its score · 1 test
Beats HappyHorse 1.1, HappyHorse 1.0, Wan 2.2 I2V Plus here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| VGI-Bench visual reasoning (I2V) | 44.0% ✎ | ▄▄···▄▄█·█·▄ beats 6/8 | 38% |
other tests2 not in a task · 55% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SWITCH interactive-interface I2V (VLM-judge overall) | 37.4% ✎ | ███··█·····█ beats 2/4 | 41% |
| MechVerse mechanical-motion I2V (WorldModelBench-judge sum) | 76.6% ✎ | ▂···▂······█ beats 5/5 | 13% |
$ where to run it
Through an API
api · Kling AI · $0.084/sec · kling.ai ↗ · price sourceUSD per second of generated video, 720P without native audio, Kling developer API list price. Product: Kling Video Generation API. API model: Kling 3.0. Settings: resolution=720P, native audio=off. Price basis: 0.6 units/s × $0.14 per unit (list price); Kling prints $0.084/s. With native audio 720P $0.126/s; 1080P $0.112/s without / $0.168/s with audio; 4K $0.42/s. Kling 3.0 Turbo and Omni excluded. Prepaid resource packages. Currency: USD. Kling AI developer pricing page, observed 2026-09-25.
$ how the score was worked out · SWITCH: Benchmarking Modeling… 41%, VGI-Bench: Probing Visual Int… 38%, MechVerse: Evaluating Physica… 13%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 SWITCH: Benchmarking Modeling and Handling of Tangible Interfaces in Long-horizon Embodied Scenarios (arXiv:2511.17649v5) | 1 | 41% |
| VGI-Bench: Probing Visual Intelligence in Video Generation Models | 1 | 38% |
| MechVerse: Evaluating Physical Motion Consistency in Video Generation Models (arXiv:2605.14843v1) | 1 | 13% |
| prior · pull toward the middle | 7% | |
| weighted average | = 67.7 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json