Details

Alex Rowland

By Alex Rowland

Category: Models

2 w

70

• Multimodal AI
• AI models
• Alibaba
• generative AI

Alibaba has unveiled Qwen3.5-Omni, an omnimodal AI model with text, image, audio, and video understanding. It is claimed to outperform Gemini 3.1 Pro on audio tasks and introduces a new capability: coding from spoken instructions and video input.

Details

Daniel Mercer

By Daniel Mercer

Category: Models

2 w

57

• Google
• Generative video
• DeepMind

Google DeepMind has introduced Veo 3.1 Lite, its most affordable video generation model to date. The new model costs less than half the price of Veo 3.1 Fast while, according to Google, maintaining comparable generation speed. However, the company has not disclosed detailed qualitative differences between Veo 3.1, Fast, and Lite.

Details

Chris Borden

By Chris Borden

Category: Industry

2 w

71

• Artificial intelligence
• generative AI

According to an extensive Rolling Stone investigation, AI generators have already become a fixed part of professional music production. Top producers and songwriters are using the technology behind the scenes.

Details

Alex Rowland

By Alex Rowland

Category: Robotics

2 w

53

• History of Technology
• Big Tech

From March 25 to 29, Beijing hosted the ZGC Forum, where developers showcased a wide range of brain–computer interface (BCI) products — from specialized chips to rehabilitation systems. The technologies attracted significant attention from visitors, according to China.org.

Details

Chris Borden

By Chris Borden

Category: Models

2 w

65

• Artificial intelligence
• generative AI

Suno has introduced its new AI music model v5.5 along with features designed to showcase musical individuality

Details

Alex Rowland

By Alex Rowland

Category: Platforms

2 w

58

• Gemini
• AI models
• Artificial intelligence

Google has introduced a so-called “Agent Skill” for the Gemini API designed to close the knowledge gap language models face when SDKs change rapidly. The issue is that AI models, once trained, are unaware of their own updates or the latest best practices. The new skill provides coding agents with up-to-date information on models, SDKs, and example code. In tests across 117 tasks, the success rate of the best model (Gemini 3.1 Pro Preview) increased from 28.2% to 96.6%

Details

Daniel Mercer

By Daniel Mercer

Category: Platforms

3 w

54

• AI Infrastructure
• Meta

Arm Holdings has introduced its own data center chip optimized for AI inference

Details

Daniel Mercer

By Daniel Mercer

Category: Industry

3 w

55

• OpenAI
• Sora
• AI industry
• Generative video

OpenAI is shutting down its Sora app along with the API. Disney is now pulling out of the partnership the two companies had signed only in December.

Details

Daniel Mercer

By Daniel Mercer

Category: Models

3 w

68

• AI Agents

The research team behind Molmo has released a web agent that can operate websites using nothing but screenshots of the screen. Despite their compact size, the models in some cases outperform larger proprietary systems.