Open-source Sora 2, realtime AI video, new DeepSeek, Nanobanana upgrades, Claude 4.5, realtime TTS
This week saw the release of several groundbreaking AI models, including video generators with transparency, real-time 3D head creation, and a new open-source text-to-speech model, alongside updates from major AI developers like Enthropic and Alibaba.
MAIN POINTS FROM TRANSCRIPT
- Sora 2 and a new open-source competitor were released, featuring video and audio generation.
- One Alpha by Alibaba generates videos with transparency, handling complex objects like hair and translucent items.
- A new real-time text-to-speech model, Canyt, offers low latency and runs on consumer-grade GPUs.
- Enthropic released Claude 4.5, and updates came from Nano Banana, GLM, and DeepSeek.
TAKEAWAYS
- AI advancements this week include models capable of real-time video editing and 3D head creation.
- One Alpha's transparency feature allows seamless video layering over different backgrounds.
- Canyt's real-time speech generation is efficient, requiring minimal VRAM for operation.
- The AI community continues to innovate with open-source releases accessible on platforms like HuggingFace and GitHub.