AI scientist, DNA editors, AI NPCs, new Qwen, open-source robots, new video editors: AI NEWS
New multimodal video editing models and advanced alignment techniques like Flash GRPO directly expand the toolkit for professional AI video creators.
What this lesson covers
This news roundup covers recent advancements including ByteDance's unified multimodal model Lance for video and image generation, and Apple's Lito 3D model generator. It also highlights Flash GRPO for video alignment and reactive game world models.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 01:04 ↗
ByteDance released Lance, a 3 billion parameter unified multimodal model capable of text-to-video and video editing.
- 01:56 ↗
Lance supports sequential video editing through multiple turns, allowing users to modify subjects and backgrounds step by step.
- 03:43 ↗
Apple introduced Lito, a 3D model generator using surface light field tokenization to capture view-dependent appearance.
- 05:01 ↗
Flash GRPO optimizes video alignment against human preferences by sampling a single time step efficiently.
Workflow outlined in the video
- Explore open-source multimodal models like Lance for advanced sequential video editing and multi-turn prompting workflows.
- Review Apple's Lito 3D model generator for new approaches to incorporating 3D assets into virtual production environments.
Before you use this workflow
These notes describe the source video at its publication date. Model access, pricing, connectors and interfaces may have changed. Check the original source and the provider’s current documentation before installing an add-on, connecting an account or spending credits.
ReelStack has not independently tested this workflow. Preview one representative shot and check motion, continuity and output quality before applying it to a full production. No result, cost saving or model capability is guaranteed.
How to interpret the numbers
93,819 views captured 23 October 2026. ReelStack recommendation score: 31/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 45% performance: 0.80× views vs 10 other channel videos observed at a similar age. View-evidence factor 99% (views / (views + 1,000))
- 25% freshness: 0/100 with a 14-day half-life
- 20% momentum: 3 views/day vs 12 channel baseline (5 comparable uploads), with the same view-evidence factor
- 10% engagement: 73/100; likes + 4× comments, smoothed with a 500-view neutral prior