Kuaishou headquarters, Building 1, No. 6 Shangdi West Road, Haidian District, Beijing, China

Kling 3.0 specs as launched

Spec	Value
Maker	Kuaishou (Kling AI)
Released	5 February 2026, early access for Ultra subscribers first (CineD, Kuaishou release)
Max clip length	15 seconds (3 to 15 s selectable on fal)
Native audio languages	English, Chinese, Japanese, Korean, Spanish, with American, British and Indian accents
Multi-shot

克林3.0发布 · Chronopin

Kling 3.0 API price on fal

Spec	Value
3.0 Pro, audio off	$0.112 per second (USD)
3.0 Pro, audio on	$0.168 per second (USD)
3.0 Pro, audio with voice control	$0.196 per second (USD)
3.0 Standard, audio off	$0.084 per second (USD)
3.0 Standard, audio on	$0.126 per second (USD)
3.0 Standard, audio with voice control	$0.154 per second (USD)

Prices as listed on fal.ai endpoint pages, read 24 Oct 2026. Standard: https://fal.ai/models/fal-ai/kling-video/v3/standard/text-to-video · 来源

Artificial Analysis text-to-video leaderboard

Entry	Result
Kling 3.0 Pro	Elo 1095, 12th
Kling 3.0 Standard	Elo 1089, 13th
Leader: Gemini Omni Flash	Elo 1233
Second: Wan 3.0	Elo 1229
Third: Minimax H3 Max	Elo 1227

Standings as of September 2026 as reported by Tech Times quoting Artificial Analysis (not read on Artificial Analysis's own JS page); vote count not published. · 来源

Kling 3.0 vs Kling 2.6

Spec	Kling 2.6	Kling 3.0
Max clip length	10 seconds	15 seconds
Native audio languages	English and Chinese	English, Chinese, Japanese, Korean, Spanish

2.6 figures from https://macaubusiness.com/kling-ai-launches-video-2-6-with-native-audio-generation-redefining-the-ai-video-creation-workflow/ · 来源

Video model rubric2026 criteria

Estimate from the version's published specs and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (Kling 3.0, 5 February 2026). Resolution 4: official tiers are 720p and 1080p (Artificial Analysis lists both); CineD reported native 4K at launch but ScriptByAI dates native 4K for Video 3.0 to 23 April 2026, so 4K is not counted. Duration 15 s (4). Audio 5: native speech in English, Chinese, Japanese, Korean and Spanish with accents and multi-character dialogue. Control 4: text, image, reference-video and multi-image references with element and voice binding and multi-shot storyboards (Omni), but not 50 references. Speed null: no figure published (a reseller blog calls it slower than 2.5 Turbo without numbers). Realism 4: CineD and Blue Lightning TV praise photorealism and continuity; The Decoder's launch story is announcement coverage, and Tech Times notes it ranked only 12th on the Artificial Analysis text-to-video arena by September 2026.^[1]

Reviews

Kling 3.0 AI Video Model Introduced – Native 4K, Enhanced Photorealism, Multi-Shot Sequencing, and Integrated Audio · cined.com · 2026年2月5日Early creator feedback highlights significantly improved photorealistic quality compared to previous versions
Kling 3 Review (2026): Is This the Best AI Video Generator? Multishot + Native Audio Tested - Blue Lightning · bluelightningtv.com · 2026年2月5日Minor color drift popped up in places, but overall continuity and realism were the best of the set.
Chinese AI video model Kling 3.0 takes another step toward usable creative assets · the-decoder.com · 2026年2月4日Key features include improved consistency for characters and elements, video production with 15-second clips and better control, and customizable multi-shot recording.
Kling 4.0 Promises 30-Second 4K Video: Broadcast Spec Is Coming in October, Not Today · techtimes.com · 2026年9月30日Kling 3.0 ranked 12th on the Artificial Analysis text-to-video leaderboard

情绪走势正面

口碑倾向

情绪走势
Chinese AI video model Kling 3.0 takes another step toward usable creative assets · 2026年2月4日	0.40
Kling 3.0 AI Video Model Introduced – Native 4K, Enhanced Photorealism, Multi-Shot Sequencing, and Integrated Audio · 2026年2月5日	0.60
Kling 3 Review (2026): Is This the Best AI Video Generator? Multishot + Native Audio Tested - Blue Lightning · 2026年2月5日	0.40
Kling 4.0 Promises 30-Second 4K Video: Broadcast Spec Is Coming in October, Not Today · 2026年9月30日	-0.20

参考资料 5可信度 79%总体可信度: 79%该图钉的来源和参考资料对其日期的支持程度。包括来源在内的 5 份资料对图钉开始和结束时间支持程度的加权平均；资料每比最新的一份旧 180 天，权重减半显示所有可信度不低于 75% 的图钉

第一项始终是图钉的来源。总体可信度是各份资料对上方所用开始和结束时间支持程度的加权平均；资料每比最新的一份旧 180 天，权重减半。

[1]90%
finance.yahoo.com/news/kling-ai-launches-3-0-080000425.html
finance.yahoo.com· 发布于 2026年10月24日· 开始 2026年2月5日 ✓· 占评分 20%
美通社发布日期为北京，2026 年 2 月 5 日（“今天宣布推出 Kling 3.0 车型”）； Decoder 于 2 月 4 日（美国时间）报道了 Ultra 抢先体验，开发者 API 于 2 月 25 日报道。
[2]85%
Chinese AI video model Kling 3.0 takes another step toward usable creative assets
the-decoder.com· 添加于 2026年10月24日· 开始 2026年2月4日· 占评分 20%
4 February 2026: an 'all-in-one creative engine' with better character consistency, 15-second clips and customizable multi-shot recording; Ultra-only early access, no API or documentation yet.^[1]
[3]75%
Kling 3.0 AI Video Model Introduced – Native 4K, Enhanced Photorealism, Multi-Shot Sequencing, and Integrated Audio
cined.com· 添加于 2026年10月24日· 开始 2026年2月5日· 占评分 20%
5 February 2026: reports native 4K, 15-second multi-shot clips and synchronized audio and early creator feedback of 'significantly improved photorealistic quality' (its 4K claim conflicts with the later 23 April 4K date).^[1]
[4]75%
Kling AI Timeline: Model Release Dates & History
scriptbyai.com· 添加于 2026年10月24日· 开始 2026年2月5日· 占评分 20%
Dates the 3.0 family to 5 February 2026, API access to 25 February and native 4K for Video 3.0 to 23 April 2026.^[1]
[5]70%
Kling 4.0 Promises 30-Second 4K Video: Broadcast Spec Is Coming in October, Not Today
techtimes.com· 添加于 2026年10月24日· 开始 2026年9月30日· 占评分 20%
30 September 2026: notes Kling 3.0 ranked 12th on Artificial Analysis's text-to-video leaderboard by then, not fourth as some coverage claimed.^[1]

建议更正

有遗漏或错误吗？用你自己的话说明：能佐证此图钉的链接、不同的开始或结束日期及理由，或缺失、有误的信息。AI 会对照此图钉的来源进行核实，搜索更好的来源，并添加任何支持你说法的页面。图钉自身的来源仍然最重要。AI 也会查看图片：显示的是别的东西或显示效果差的图片会被移到后面或替换。

AI发布于 2026/10/242026/10/24 16:42@AIDesk

去时间线

Kuaishou

开始于 2026/02/05已确认可信度 79%

美通社发布日期为北京，2026 年 2 月 5 日（“今天宣布推出 Kling 3.0 车型”）； Decoder 于 2 月 4 日（美国时间）报道了 Ultra 抢先体验，开发者 API 于 2 月 25 日报道。^[1]

开始: 2026年2月5日可信度 90%来自来源 · 可能为 2026/2/4 – 2026/9/30

克林3.0发布

Name: 克林3.0发布
Author: Kuaishou

由原文自动翻译

原标题: Kling 3.0 Released

Kling AI 推出 3.0 系列，包括 Video 3.0 和 Video 3.0 Omni，包含 15 秒剪辑、带口音和方言的本地多语言音频、多重镜头叙事和视频参考角色，首先向 Ultra 订阅者推出。

Kling AI 于 2026 年 2 月 5 日宣布推出 3.0 系列：Video 3.0、Video 3.0 Omni、Image 3.0 和 Image 3.0 Omni，基于 O1 和 2.6 模型以及公司的多模态视觉语言框架构建，优先考虑 Ultra 订阅者
视频3.0将剪辑延长至15秒，添加了带有美国、英国和印度口音的英语、中文、日语、韩语和西班牙语的母语语音，以及每个人都说不同语言的多角色场景
它理解多镜头指令（镜头-反向镜头、横切），并且 Omni 接受参考视频以在场景中传递角色的外观和声音，以及每个镜头的故事板
解码器描述了一种“一体化的创意引擎”^[2]CineD 引用了早期创作者的反馈，称其具有更好的照片真实感，尽管其原生 4K 声明早于 4 月 23 日实际 4K 发布^[3][4]
API 于 2 月 25 日发布，3.0 Turbo 于 6 月 17 日发布^[4]; 到 2026 年 9 月，它在 Artificial Analysis 的文本到视频领域排名第 12 位^[5]

@AIDesk