New top AI video models, Google dominates, top 3D model generator, Grok 4.1, GPT-Codex-Max: AI NEWS
The release of advanced open-source depth and 3D reconstruction models like Depth Anything 3 and SAM 3D provides indie filmmakers with powerful local tools for spatial computing and camera tracking workflows.
What this lesson covers
This AI news roundup covers major open-source releases including Depth Anything 3 for 3D depth and geometry estimation, Meta's SAM 3 for object segmentation, and SAM 3D for human mesh and object 3D generation. The video also highlights updates to HunyuanVideo 1.5, Grok 4.1, Gemini 3, and GPT-5.1-Codex-MAX.
Key takeaways from the creator
AI-extracted notes, not independently verified product claims. Timestamp links let you check each point in the original video.
- 01:13 ↗
Depth Anything 3 can generate a coherent 3D map, depth, and geometry of a large space using a few reference photos or chaotic video feeds.
- 01:41 ↗
The largest version of Depth Anything 3, DA3 Nested Giant Large, is only 1.4 billion parameters and can run locally on consumer-grade GPUs.
- 03:49 ↗
Meta's Segment Anything Model 3 (SAM 3) allows users to detect, segment, and track objects in images and video using text prompts or click inputs.
- 05:48 ↗
SAM 3D converts specified objects and human bodies from a single image into accurate 3D models and body meshes, handling irregular poses like yoga postures.
- 08:40 ↗
Tencent's HunyuanVideo 1.5 was released as a state-of-the-art open-source video generator alongside other industry updates.
Workflow outlined in the video
- Visit the Depth Anything 3 GitHub repository to review installation instructions and hardware requirements for local execution.
- Test Meta's SAM 3 on reference video footage for object tracking and segmentation tasks.
- Explore the SAM 3D online demo to experiment with converting 2D character images into 3D meshes.
- Evaluate the memory footprint of the DA3 Nested Giant Large model against your local consumer GPU specifications.
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
108,346 views captured 23 October 2026. ReelStack recommendation score: 41/100. Formula: radar-v5. This is ReelStack’s calculation, not a YouTube rating or a measure of factual accuracy.
Show the calculation
- 45% performance: 1.06× 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 4 channel baseline (6 comparable uploads), with the same view-evidence factor
- 10% engagement: 86/100; likes + 4× comments, smoothed with a 500-view neutral prior