■ The Big Story

Google cuts the price of making short AI videos

Google DeepMind says Gemini Omni 1.2 Flash can create and edit videos up to 15 seconds long, up from 10 seconds with Gemini Omni 1.1. It is generally available through the Gemini API and Gemini for business. The company says 720p video costs $0.07 per second, 30% less than Gemini Omni 1.1. Marketing teams now have a clearer cost for testing short video variations.

Google DeepMind says teams can choose clip lengths in one-second steps from three to 15 seconds. That makes a difference when an ad needs a shorter opening or a product explainer needs several cuts: teams can pay for the length they need rather than defaulting to the longest clip. Separately, Anthropic says Personal Finance in Claude is rolling out in the US, letting users ask about connected bank and card accounts through read-only access.

Sources: [Google DeepMind] [Anthropic]

■ Frontier Watch

■ The Street Says

The sharp selloff has put the cost of building AI infrastructure under scrutiny, even as AI companies pitch larger roles for their products inside businesses.

■ Repo of the Day

■ Two Steps Ahead

Over the next 30–90 days, marketing teams using Gemini Omni 1.2 Flash can ask agencies to separate concept development and final finishing from charges for generating draft clips. The price change makes draft production easier to compare; it does not erase the value of editing, approvals or distribution.

For security procurement, GPT-6 Cyber’s restricted release makes access terms more important than a headline capability claim. Ask testing providers whether approved Daybreak Red access is included in your contract. Meanwhile, Mistral’s Studio gives businesses with data-location requirements a concrete deployment alternative to put beside cloud-only proposals.

If an internal data assistant is being scoped this quarter, NVIDIA / auto-ontology points to a different purchasing question: how will staff trace an answer to company records? Make that a requirement before comparing assistants on how fluent their responses sound.