Weekly AI News Roundup: New Video Models, Business Tools, and Agent Development
This week, Google rolls out longer video generation in Google Vids, OpenAI introduces a benchmark for biological labs, and Mistral AI launches a platform for building enterprise agents. Here are the key updates for businesses.
This roundup was prepared automatically from official sources. Each story links to its original announcement.
Google Vids introduces longer AI-generated videos for free
Google Vids now allows generating 15-second video clips using the Omni 1.2 Flash model. The feature is available for free to Google account holders or Google Workspace subscribers, with a limit of three clips per month.
Businesses can use this tool to quickly create short video content like ads or product explainers without relying on creative teams or external vendors.
Gemini Omni 1.2 Flash: longer videos, lower costs, and improved consistency
Google releases Gemini Omni 1.2 Flash, enabling 15-second video generation, multi-turn editing, and improved visual consistency. The cost of video generation has been reduced by 30%.
Companies can create consistent video campaigns while maintaining brand elements like colors and style, making it easier to scale ad content production.
OpenAI introduces BioBasicsBench for biological labs
OpenAI launches BioBasicsBench, a 100-task benchmark measuring AI agents' ability to interpret experimental data and handle everyday processes in biological labs.
Biotech and pharmaceutical companies can use this benchmark to evaluate and deploy AI agents that streamline lab workflows, reducing time and costs for experiments.
Gemini turns sketches into functional apps
Google introduces the Canvas tool in Gemini, enabling app code generation from sketches. Users can upload a drawing, and the system automatically creates a functional prototype.
Businesses can accelerate app prototyping, allowing non-technical teams to quickly turn ideas into functional solutions without writing code from scratch.
Mistral AI launches platform for building enterprise agents
Mistral AI releases Mistral Studio, a platform for building and deploying enterprise agents, workflows, and business applications. The new Mistral Large 4 (Le Chonk) model offers advanced coding and document analysis capabilities.
Enterprises can automate business processes like document handling or data analysis without building solutions from scratch, speeding up AI adoption in the company.
This week’s takeaway
New AI tools and models offer businesses opportunities to automate workflows, create video content, and deploy agents without requiring deep technical expertise. Keeping track of these developments will help companies apply them effectively.