KuaishouVideo model
Kling 2.6 Pro
In image to video, Kling 2.6 Pro is in place 16 of 21, with text only in place 18. A 5 s clip costs about $0.70.
1,294Arena ratinghigher is better
1,216Arena ratinghigher is better
$0.70$0.14 per secondlower is better
n/alongest single cliphigher is better
29out of 100same as last month
Summary
At $0.14 per second of video it is fairly expensive; a typical video model costs $0.12.
Kling 2.6 Pro was Kuaishou’s high-resolution Kling generation for API and web access. The proprietary model generated up to 1080p and included native audio in the clip.
Kuaishou explicitly names the 2.6 series as a foundation for Kling 3.0 Pro. Storyboard and reference-video controls belong to the later 3.0 generation. Kling 2.6 Pro remained at 1080p.
Kling 2.6 Pro benchmarks
Image to video arena
How people rate generated videos in blind comparisons, Elo points · higher is better
20 of 21 models
as of September 22, 2026 · Source: LMArena
Text to video arena
How people rate generated videos in blind comparisons, Elo points · higher is better
20 of 23 models
as of September 22, 2026 · Source: LMArena
Show all 21 individual results from 6 sources
Page 1 of 2
Published benchmarks
LMArena Text to Video
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of November 5, 2026Scale: 0 Elo - 1,516 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 23
- Sample
- 73,362
LMArena Image to Video
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of November 5, 2026Scale: 0 Elo - 1,495 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 21
- Sample
- 202,113
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
vidbenchmark blind video benchmark (Show measurement, test conditions, and source)
- Source value
- 1090
- Score
- 1090
- Metric
- Blind same-prompt human pairwise Elo
- Unit
- elo
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 76
- Score
- 76
- Metric
- Prompt adherence score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 72
- Score
- 72
- Metric
- Motion realism score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 71
- Score
- 71
- Metric
- Character consistency score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 71
- Score
- 71
- Metric
- Temporal coherence score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 69
- Score
- 69
- Metric
- Lip-sync accuracy score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
vidbenchmark eight-axis model scores (Show measurement, test conditions, and source)
- Source value
- 66
- Score
- 66
- Metric
- Native audio quality score (0-100 source score)
- Unit
- points
- Benchmark version
- vidbenchmark dataset snapshot (2026-08-04)
- Category
- video-generation-quality
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- vidbenchmark
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact source model entry kling-2-6-pro (Kling 2.6 Pro). Methodology describes blind same-prompt human pairwise voting and a shared duration/resolution target; the exact global target and this row’s realized output setting are not stated. The dataset lists maxDurationSeconds=10, maxResolution=1080p, which are model limits, not the proven test settings. Data snapshot is 2026-08-04.
- Context
- The website claims weekly reruns, but the downloaded dataset says updatedAt=2026-08-04. No sample count, vote count, confidence interval, or test resolution is published in this JSON. Do not treat the model maximum as the tested resolution, or compare this unspecified task directly with T2V/I2V-specific boards. The methodology page discloses affiliate revenue.
RBench average score (Show measurement, test conditions, and source)
- Source value
- 0.534
- Score
- 0.534
- Metric
- Avg across nine robot/task indicators
- Unit
- ratio
- Benchmark version
- RBench Table 2
- Category
- image-to-video
- Direction
- Higher is better
- Sample
- 650
- Source type
- independent-evaluator
- Evaluator
- RBench, Rethinking Video Generation Model for the Embodied World
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Kling 2.6 Pro; image-to-video for embodied robotics; 1920 × 1080; 5 s; 24 fps.
- Context
- Robotics-oriented image-to-video evaluation, not general video quality; no audio score is reported.
RBench per-category score (Show measurement, test conditions, and source)
- Source value
- 0.529
- Score
- 0.529
- Metric
- RBench category score
- Unit
- No unit provided
- Benchmark version
- RBench Table 2, arXiv:2601.15282v1
- Category
- task-manipulation
- Direction
- No direction published
- Sample
- 50
- Source type
- independent-evaluator
- Evaluator
- RBench original paper
- Status
- active
- Retrieved at
- 2026-10-19
- Methodology
- Independent evaluation. RBench Table 2, exact “Kling 2.6 pro” row, Manipulation category. Closed-source models use official APIs following developer-recommended invocation. The paper generates three videos per image-text sample and averages them before scoring. The test set reports 50 examples for this category. The table does not state a unit or confidence interval for this displayed category score.
RBench per-category score (Show measurement, test conditions, and source)
- Source value
- 0.598
- Score
- 0.598
- Metric
- RBench category score
- Unit
- No unit provided
- Benchmark version
- RBench Table 2, arXiv:2601.15282v1
- Category
- task-spatial
- Direction
- No direction published
- Sample
- 50
- Source type
- independent-evaluator
- Evaluator
- RBench original paper
- Status
- active
- Retrieved at
- 2026-10-19
- Methodology
- Independent evaluation. RBench Table 2, exact “Kling 2.6 pro” row, Spatial category. Closed-source models use official APIs following developer-recommended invocation. The paper generates three videos per image-text sample and averages them before scoring. The test set reports 50 examples for this category. The table does not state a unit or confidence interval for this displayed category score.
Recorded measurement details
LMArena Text to Video: LMArena Text to Video (Show measurement, test conditions, and source)
- Source value
- 1216
- Score
- 1216
- Metric
- elo-rating
- Unit
- Elo
- Task
- lmarena-text-to-video
- Task label
- LMArena Text to Video
- Direction
- Higher is better
- Source rank
- 29
- Sample
- 73362
- Lower confidence bound
- 1209
- Upper confidence bound
- 1223
- Retrieved at
- 2026-11-05
LMArena Image to Video: LMArena Image to Video (Show measurement, test conditions, and source)
- Source value
- 1294
- Score
- 1294
- Metric
- elo-rating
- Unit
- Elo
- Task
- lmarena-image-to-video
- Task label
- LMArena Image to Video
- Direction
- Higher is better
- Source rank
- 26
- Sample
- 202113
- Lower confidence bound
- 1286
- Upper confidence bound
- 1302
- Retrieved at
- 2026-11-05
Page 1 of 2
Video sample from our test
This video exists as a documented run for the exact model version. Playback is muted.
Motion and physical consistency
This sample tests Anatomy, object consistency, camera tracking, rain, and reflections.
Prompt and settings
Generate exactly one continuous five-second 16:9 cinematic video. A fictional bicycle courier in a mustard-yellow raincoat rides a red city bicycle from left to right across a rain-wet Hamburg street at blue hour for the entire clip. The camera tracks her smoothly at waist height. Both hands remain visible on the handlebars, the bicycle geometry stays physically correct, and a transparent umbrella fixed to the bicycle moves naturally in the wind. Wet-street reflections and raindrops follow realistic physics. No cuts, no slow motion, no text, no logos, no subtitles, and no audio.
Generated through fal.ai with model fal-ai/kling-video/v2.6/pro/text-to-video. Archived settings were 16:9, one request, first output, requested 5 s, seed provider random, request {"prompt":"Generate exactly one continuous five-second 16:9 cinematic video. A fictional bicycle courier in a mustard-yellow raincoat rides a red city bicycle from left to right across a rain-wet Hamburg street at blue hour for the entire clip. The camera tracks her smoothly at waist height. Both hands remain visible on the handlebars, the bicycle geometry stays physically correct, and a transparent umbrella fixed to the bicycle moves naturally in the wind. Wet-street reflections and raindrops follow realistic physics. No cuts, no slow motion, no text, no logos, no subtitles, and no audio.","duration":"5","aspect_ratio":"16:9","generate_audio":false}, configured price estimate $0.35.
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
0 to 100, higher is more popular
20 of 98 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 5, 2026
Kling 2.6 ProGrok Imagine Video 1.5HunyuanVideo 1.5Dreamina Seedance 2.0 720pWan 2.7
Kling 2.6 Pro compared
Compare with another model
Predecessor and successor
The most popular comparisons
- 1Gemini Omni Flash vs. Kling 2.6 Pro
- 2FLUX.3 Video vs. Kling 2.6 Pro
- 3Dreamina Seedance 2.5 720p vs. Kling 2.6 Pro
- 4Dreamina Seedance 2.0 720p vs. Kling 2.6 Pro
- 5MiniMax H3 vs. Kling 2.6 Pro
- 6Sora 2 Pro vs. Kling 2.6 Pro
- 7Sora 2 vs. Kling 2.6 Pro
- 8Veo 3.1 vs. Kling 2.6 Pro
- 9Veo 3.1 Fast vs. Kling 2.6 Pro
Show all 47 comparisons
- 10Grok Imagine Video 1.5 vs. Kling 2.6 Pro
- 11Grok Imagine Video vs. Kling 2.6 Pro
- 12Wan 2.7 vs. Kling 2.6 Pro
- 13Wan 2.6 vs. Kling 2.6 Pro
- 14Runway Gen-4.5 vs. Kling 2.6 Pro
- 15Kling 2.6 Pro vs. Vidu Q3 Pro
- 16Kling 2.6 Pro vs. PixVerse V5.6
- 17Kling 2.6 Pro vs. Hailuo 2.3
- 18Kling 2.6 Pro vs. LTX-2 19B
- 19Kling 2.6 Pro vs. Wan 2.2 A14B
- 20Kling 2.6 Pro vs. HunyuanVideo 1.5
- 21Kling 2.6 Pro vs. Pika 2.2
- 22Kling 2.6 Pro vs. Gemini Omni 1.1 Flash
- 23Kling 2.6 Pro vs. Muse Video
- 24Kling 2.6 Pro vs. Wan 3.0
- 25Kling 2.6 Pro vs. Wan 3.0 Prime
- 26Kling 2.6 Pro vs. HappyHorse 1.0
- 27Kling 2.6 Pro vs. HappyHorse 1.1
- 28Kling 2.6 Pro vs. PixVerse V6
- 29Kling 2.6 Pro vs. PixVerse C1
- 30Kling 2.6 Pro vs. Seedance 2.0 Fast
- 31Kling 2.6 Pro vs. Seedance 2.0 Mini
- 32Kling 2.6 Pro vs. Vidu Q3 Turbo
- 33Kling 2.6 Pro vs. Vidu Q2
- 34Kling 2.6 Pro vs. Vidu Q1
- 35Kling 2.6 Pro vs. Kling 4.0
- 36Kling 2.6 Pro vs. Kling 3.0 Standard API
- 37Kling 2.6 Pro vs. Kling Video O1
- 38Kling 2.6 Pro vs. Hailuo 2.3 Fast
- 39Kling 2.6 Pro vs. Wan 2.5 Preview
- 40Kling 2.6 Pro vs. Wan 2.6 I2V Flash
- 41Kling 2.6 Pro vs. Wan 2.2 TI2V 5B
- 42Kling 2.6 Pro vs. Wan 2.2 I2V A14B
- 43Kling 2.6 Pro vs. HunyuanVideo 1.0 T2V
- 44Kling 2.6 Pro vs. HunyuanVideo I2V 1.0
- 45Kling 2.6 Pro vs. Ray 3.2
- 46Kling 2.6 Pro vs. Grok Imagine Video 1.5 Lite
- 47Kling 2.6 Pro vs. LTX-2.5
All specifications
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Unknown
- no source
- Watermarking technique
- Not known
- no source
Licence and access
Prices
No Kuaishou subscription includes this model. Through other providers from €6.83 a month (Poe Basic, with limited points). 12 plans in Germany.
| Plan | Monthly price | Note |
|---|---|---|
| Poe Basic | €6.83 | with limited credit |
| Runway Standard | ≈ €15.90 | with limited credit |
| Recraft Pro | ≈ €21.20 | with limited credit |
| Recraft Teams | ≈ €23.33 | with limited credit, for teams, per seat |
| Poe Plus | €27.36 | with limited credit |
| Runway Pro | ≈ €37.11 | with limited credit |
| Poe Pro | €68.41 | with limited credit |
| Runway Team | ≈ €73.16 | with limited credit, for teams, per seat |
| Runway Max | ≈ €100.72 | with limited credit |
| Poe Advanced | €136.84 | with limited credit |
| Poe Max | €342.11 | with limited credit |
| Runway Enterprise | n/a | for enterprises |
Frequently asked questions about Kling 2.6 Pro
How good is Kling 2.6 Pro?
Rank 18 of 23 in the text-to-video arena rating. With a start image: rank 16 of 21.
LMArena, as of Sep 22, 2026
How much does a video with Kling 2.6 Pro cost?
About $0.70 for a 5 s clip, $0.14 per second of video. That is fairly expensive.
Where can I use Kling 2.6 Pro?
Not in a Kuaishou subscription. Including other providers that makes 12 plans in Germany.
Subscription comparison, prices in Germany incl. VAT
At Kuaishou
No Kuaishou subscription includes this model.
Through other providers
12 plans from 3 providers, from €6.83 a month (Poe Basic), with limited credits.
Sources and data date
Show all 19 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
| Evaluator | Test series | Composite indices | Recorded active measurements |
|---|---|---|---|
| LMArena | 0 | 0 | 2 |
Without vendor claims and historical or disputed results; those stay labelled in the single values. 2 measurements are not yet assigned to a test series.
| Type | Evidence and data date |
|---|---|
| Research status | Research date October 16, 2026. 13 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | wiro-video-models-exact-row Note The inspected Wiro leaderboard has no exact model-version route. Nearby variants were not transferred. |
| Not found in the inspected sources | arena-video-edit-exact-row Note The inspected Video Edit table has no exact model-version row. T2V and I2V values were not reused. |
| Not found in the inspected sources | vabench-exact-model-row Note VABench has no exact model row. Sora 2 was not transferred to Sora 2 Pro or other models. |
| Not found in the inspected sources | artificial-analysis-t2v-exact-row Note No exact model version or resolution was found in the inspected Artificial Analysis table. |
| Not found in the inspected sources | artificial-analysis-i2v-exact-row Note No exact model version or resolution was found in the inspected Artificial Analysis table. |
| Unresolved | vbench-current-public-leaderboard Note The current VBench leaderboard is embedded in a dynamic Hugging Face iframe and was not retrievable; current scores remain unresolved. |
| Unresolved | vidbenchmark-performance-axes Note The source publishes speed-to-first-frame and cost-efficiency axes, but its inspected methodology does not define them as measured performance results or disclose units, direction, or a scoring formula. They are omitted. Model limits, maximum resolution, and list prices are not quality measurements. |
| Additional source | Kuaishou · Kling API model documentation · Editorial description reviewed October 2, 2026 |
| Additional source | Video pricing |
| Additional source | fal.ai |
| Additional source | LMArena (retrieved November 5, 2026)12 |
| Additional source | vidbenchmark (retrieved October 16, 2026) |
| Additional source | RBench, Rethinking Video Generation Model for the Embodied World (retrieved October 17, 2026; October 19, 2026) |
| Additional source | Arena Video Image-to-Video (retrieved September 21, 2026; October 16, 2026; October 28, 2026) |
| Additional source | Arena Video Text-to-Video (retrieved September 21, 2026; October 16, 2026; October 28, 2026) |
| Additional source | Model metadata (retrieved October 16, 2026)123 |
| Additional source | Kuaishou Kling 3.0 launch announcement · Editorial description reviewed October 2, 2026 |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Ollama: cumulative downloads for the popularity index |
| Additional source | Replicate: cumulative runs for the popularity index |
| Additional source | Google Trending Now: limited feed selection for Germany and the US |
| Additional source | Creative Commons Attribution 4.0 |