Artificial Intelligence

Mar 31, 2026

Google Deepmind is launching Veo 3.1 Lite, its most affordable video generation model yet. It costs less than half the price of Veo 3.1 Fast but matches its speed, according to Google. The company doesn't specify quality differences between the three tiers.

Veo 3.1 Lite supports text-to-video and image-to-video at 720p and 1080p in portrait and landscape formats, with clips of 4, 6, or 8 seconds. Pricing starts at $0.05 per second for 720p. Starting April 7, Google is also dropping Veo 3.1 Fast prices.

Pricing (per second in USD)	Veo 3.1 Lite	Veo 3.1 Fast	Veo 3.1
720p	$0.05	$0.15 ($0.10 from 4/7)	$0.40
1080p	$0.08	$0.15 ($0.12 from 4/7)	$0.40
4K	-	$0.35 ($0.30 from 4/7)	$0.60

Veo 3.1 Lite is available now through the Gemini API and Google AI Studio, with more video news for developers coming soon. Full details are in the developer documentation.

Google's announcement comes right after OpenAI revealed it's shutting down Sora. That leaves Google facing serious video generation competition primarily from China, especially Alibaba's Seedance 2.0, which delivers higher quality but comes with copyright concerns.

Comment Source: Google

Maximilian Schreiner

Mar 31, 2026

Frontier Radar

Frontier Radar #2: Why AI productivity gets lost between benchmarks and the balance sheet

Jonathan Kemper

Mar 31, 2026

AI research

Qwen3.5-Omni learned to write code from spoken instructions and video without anyone training it to

Alibaba's advertising graphic shows two teddy bears in traditional Chinese clothing. The bear on the left is sitting at a desk in front of a monitor and represents Qwen3.5-Omni-Plus with functions such as SOTA Performance, Detailed Audio-Visual Captioning, Native Multimodal and Extensive Multilingual. The bear on the right is holding a smartphone and represents Qwen3.5-Omni-Plus-Realtime with Voice Control, WebSearch Tool, Voice Clone and Semantic Interruption.

Maximilian Schreiner

Mar 31, 2026

Short News

AI infrastructure company Nebius Group is building a 310-megawatt data center in Lappeenranta, Finland, close to the Russian border. The project is valued at over $10 billion and would become one of the largest AI data centers in Europe. Finnish developer Polarnode is already constructing the facility, with a phased launch planned starting in 2027.

Nebius recently signed contracts totaling more than $40 billion with Microsoft and Meta. The new data center will train AI models and run AI applications but isn't tied to a single customer. Nebius picked Finland for its low energy prices, renewable power, and cool climate, all of which help cut cooling costs. The facility would be the company's largest site outside the US and is expected to cover roughly 10 percent of Nebius' total planned capacity, according to CEO Arkady Volozh.

Comment Source: Reuters

Maximilian Schreiner

Mar 31, 2026

Short News

California Governor Gavin Newsom signed an executive order on Monday requiring companies with state contracts to implement safeguards against AI misuse. Specifically, companies must ensure their AI systems don't generate illegal content, reinforce harmful biases, or violate civil rights. To prevent misinformation, state agencies will also be required to watermark AI-generated images and videos.

The order includes a separate provision for handling federal directives: if the U.S. federal government designates a company as a supply chain risk, California will conduct its own review and potentially continue working with that vendor. This comes in the wake of the Pentagon's designation of Anthropic as a supply chain risk, which bars government contractors from using Anthropic's technology for U.S. military work.

Within 120 days, California's procurement and technology agencies are expected to develop recommendations for new AI certifications. These would let companies demonstrate compliance with responsible AI practices and public safety protections.

The executive order reinforces California's push to chart its own course on AI regulation, independent of the Trump administration, which has repeatedly tried to block independent state-level AI laws.

Comment Source: Reuters

Maximilian Schreiner

Mar 31, 2026

AI in practice

OpenAI launches a Codex plugin that runs inside Anthropic's Claude Code

Matthias Bastian

Mar 30, 2026

AI in practice

Insiders liken AI to "the Ozempic of the music industry" as hitmakers reportedly hide their generator use

The use of AI generators in professional music production is growing fast, but the industry would rather not talk about it. Top producers and songwriters are quietly embracing the technology behind the scenes, and while established creatives are finding new opportunities, an entire class of working musicians fears for its livelihood, according to extensive research by Rolling Stone.

Read full article

Comment

Matthias Bastian

Mar 30, 2026

Short News

Microsoft is making "Copilot Cowork" more widely available and launching a new AI research agent. The previously announced feature builds on Claude Cowork and lets the system handle multi-step tasks using tools, accessing and outputting files. It also includes calendar planning and daily briefings. The feature is available as part of the Frontier program.

Microsoft's "Researcher" tool now has a "Critique" function where one AI model writes a draft and a second one reviews it. It pulls from both Anthropic and OpenAI models. Microsoft says the new agent hits best-in-class deep research performance and outperforms Perplexity with Claude Opus 4.6 by 7 points. Microsoft's benchmark doesn't include a comparison with OpenAI's new GPT-5-based Deep Research, though.

There's also a new "Model Council" feature where users can compare answers from different AI models side by side to see where they agree or differ. All of these updates ship as part of Wave 3 of Microsoft 365 Copilot.

Comment Source: Microsoft Tech Blog | Copilot Cowork

Maximilian Schreiner

Mar 30, 2026

AI research

AI models confidently describe images they never saw, and benchmarks fail to catch it

Matthias Bastian

Mar 30, 2026

Short News

OpenAI's Sora app saw rapidly declining usage while costing the company around one million dollars a day, according to the Wall Street Journal. After a hyped launch, the app grew to about one million users, but that number quickly dropped to around 500,000 and never recovered.

On top of the shrinking user base, OpenAI ran into copyright issues and growing internal concerns that the cheap, low-quality engagement videos people were generating could damage the OpenAI brand. Sora proved more liability than asset. Development costs piled up too. According to the report, OpenAI canceled training runs for new video models entirely.

The real nail in Sora's coffin was increasing competitive pressure from Anthropic. OpenAI chose to redirect its limited compute toward coding, enterprise, and agent-based AI products, areas with greater long-term business value. Sora fell victim to a strategic pivot: away from complex video generation, toward the most economically promising parts of the business. The Sora team will now focus on world models for robotics. The Sora app shuts down in April, with the API following in September.

Comment Source: Wall Street Journal