Discover 30-second storytelling, up to 4K output, flexible keyframes and multimodal creative control. Kling 4.0 is scheduled for October 2026; Flash has limited early access. 光谱汇深 API access is coming soon.
Official model specifications, with 光谱汇深 API access still coming soon
Kuaishou has announced Kling 4.0 for October 2026 and opened limited Flash access on September 28. This page presents the official model capabilities; 光谱汇深 availability, supported API parameters and pricing will be confirmed separately.
Updated September 28, 2026
Read the official Kling 4.0 announcementExplore the Kling 4.0 Flash previewMain-model specifications are listed below; Flash has its own limits. Availability through 光谱汇深 will be confirmed after integration.
Applications of the announced capabilities for teams preparing future Kling 4.0 workflows.
Official release plans, model differences and upcoming 光谱汇深 access.
Kling 4.0 is Kuaishou’s announced next-generation AI video model. Its published capabilities include 3–30-second generation, up to 4K output, up to 10 keyframes, multimodal references, stereo audio and video editing.
The official announcement schedules Kling 4.0 for October 2026. Kling 4.0 Flash opened limited early access on September 28, 2026. These are Kuaishou’s release plans; 光谱汇深 API availability will be announced separately.
Kling 4.0 is announced with 3–30-second generation at 720p, 1080p or 4K, plus first-and-last-frame and flexible-keyframe control. Flash lists 3–20-second generation at 720p and 8-bit SDR, with text-to-video, first-frame image-to-video and multimodal references. Main-model limits should not be assumed to apply to Flash.
For Kling 4.0, the combined total is up to 15 image, video, voice and subject reference items. Individual limits are 10 images, 5 videos totaling at most 30 seconds, and 7 subjects including at most 3 video subjects. Video-subject duration counts toward the video total. Supported 光谱汇深 API fields will be documented after integration.
Not yet. This is a preview of the announced models. 光谱汇深 will publish supported models, endpoints, parameters and availability after integration is confirmed; early access on Kling’s own platform does not mean 光谱汇深 API access is live.
光谱汇深 pricing for Kling 4.0 and Flash has not been announced. Model-specific credit prices and billing rules will be published when access is confirmed.
The official announcement marks 10-bit HDR at 1080p and 4K, and repeated video extension up to 2 minutes, as coming soon. Two minutes refers to the result of repeated extensions, not a single generation; the main model’s announced single-generation duration is 3–30 seconds.
You can submit a support ticket from the dashboard or email us at support@apimart.ai. Our team will get back to you as soon as possible.
Explore more models in the same category.
Wan 3.0 Video Prime
Wan 3.0 Prime is the Prime version of Alibaba Cloud's all-in-one Wan 3.0 video model, with general, first-and-last-frame and file or webpage modes and 2–30 second output.
MiniMax H3 Max
MiniMax H3 Max generates video from text and images at 480P, 768P or 1080P with 5–15 second clips, first/last frames, multimodal references and synced audio.
Seedance 2.0 Mini
Seedance 2.0 Mini is the lightweight version of ByteDance's Seedance 2.0 for text-to-video and image-to-video, with 480p or 720p output.
HappyHorse 1.1
HappyHorse 1.1 generates 720P or 1080P video from text, a first-frame image or 1–9 reference images. Video editing is not supported.