Kling 4.0 launches with 15-reference inputs, native 30s clips, 4K, stereo audio and API
xiaohu · x · 2026-10-22
Kuaishou's Kling team fully launched Kling 4.0 plus a lighter 4.0 Flash, with the API available globally. Key upgrades: up to 15 reference inputs per task (10 images, 5 video clips ≤30s, 7 subjects, voice), native 30-second generation with 10 keyframes and 8,000-token prompts, 720p/1080p/4K output with new 21:9 aspect, stereo audio and multilingual lip-sync, video editing to add/remove/swap subjects and backgrounds, plus an agent canvas with MCP and CLI access. Already integrated by Higgsfield, Magnific, Meitu, OpenArt, Tencent apps and others. xiaohu compiled 74 official demo videos and 80 reference images.
Related event: Kuaishou Launches Kling 4.0 with 30-Second Native Video and Global API(8 posts)→
More from Multimodal
- Dense instance segmentation via Muse Wonder + SAM 3.1 on Meta's Model API — nikhilaravi · 2026-10-26
- LichtFeld Studio v0.5.4.3 ships community-built SteadySplats, POPSpa and Voxelization — janusch_patas · 2026-10-26
- One prompt to a 2-3 min narrated short: hands-on test of Antigravity video generation — AI_Andrew · 2026-10-26
- Creator makes 10,564-frame animated short entirely with Gemini toolchain — AI_Andrew · 2026-10-26
- FLUX.3 Klein step-distilled to 4-8 steps with CFG baked in — Fit-Structure-6489 · 2026-10-26
- ostris pre-training "Primal VAE": 1-channel latents match modern 8x VAEs via patching — ostrisai · 2026-10-26