Story cards stay below the sticky dock while audio, chapters, date, and brief navigation remain accessible.
Alibaba's Wan 2.7 launched as a cloud service April 6 with a 27B MoE architecture, first-and-last-frame control, subject+voice referencing, and a novel "Thinking Mode" that reasons about prompts before rendering. Open weights are expected to hit the Wan-Video GitHub within 4-8 weeks following Alibaba's established pattern — meaning local ComfyUI deployment could land by late May or early June.
Runway Gen-4 now generates synchronized audio tracks alongside video in a single pass — matching the audio-native trend set by Kling and Seedance. New API hooks enable hybrid pipelines combining Gen-4 with Veo3, Seedance, or other platforms for mixed-model workflows.
Google's Veo 3.1 Lite is now available via the Gemini API at under 50% the cost of Veo 3.1 Fast. Consumer-side, any Google account gets 10 free video generations per month through Google Vids. New $300 Cloud credits for new accounts cover roughly 100 minutes of API video.
Kuaishou's Kling 3.0 generates up to 6 distinct camera cuts within a single clip at native 4K/60fps with synchronized audio. Subject Binding maintains character identity across all shots. Supports auto or custom storyboard modes with per-shot duration and angle control.
TikTok's C2PA-based automated detection system achieves 94.7% accuracy on synthetic faces and 87.3% on AI backgrounds. Videos flagged without proper labeling receive zero Creator Fund earnings for 14 days, even after retroactive labeling. Two violations trigger a 60% chance of shadow-restriction.
OpenAI confirmed Sora's two-stage shutdown after losing $1M/day. The consumer app and web experience went dark April 26; API access ends September 24. All user data will be permanently deleted. The market has consolidated around Veo, Kling, Wan, Seedance, and Runway.
ByteDance's unified audio-video joint generation model is now accessible via fal.ai as of April 9. Generates video with phoneme-level lip-sync in 8+ languages. Note: face generation from real faces is now blocked after Disney's cease-and-desist, so plan reference workflows accordingly.
Recent ComfyUI updates include native "Kling Text to Video with Audio" and "Kling Image to Video with Audio" nodes, an "OpenAI GPT Image" node for gpt-image-2, and 4K resolution support for Veo models. Template library expanded with production-ready workflows.
YouTube's January 2026 policy update is now fully enforced: AI-generated content remains eligible for the Partner Program (1,000 subs + 4,000 watch hours or 10M Shorts views) provided creators disclose synthetically generated media. No CPM penalty for properly labeled AI content.