Google DeepMind
·
8 months ago
Google released the stable version of Gemini 2.5 Flash-Lite, a lightweight model designed for fast, cost-efficient inference across tasks like translation and classification. The model costs $0.10 per million input tokens and $0.40 per million output tokens, with a 1 million-token context window and support for native tools including grounding with Google Search and code execution. Users can deploy the model by specifying "gemini-2.5-flash-lite" in their code, with Google retiring the preview alias on August 25th.
Google DeepMind
·
8 months ago
Eliza McNitt's short film "ANCESTRA" premiered at the Tribeca Festival using Google DeepMind's Veo video generation model combined with live-action footage, marking the first collaboration between Google DeepMind and director Darren Aronofsky's production venture. Over 200 filmmaking professionals worked alongside AI researchers to blend generated sequences—including a realistic newborn baby and interior body scenes—with traditional cinematography, using Gemini to refine prompts and Veo's new capabilities for motion matching and object insertion. The partnership established new tools for directors to control generative video output through precise motion templates and art direction, reducing production time for complex shots from hours to minutes while maintaining creative control over final sequences.
