🎬 Black Forest Labs ships FLUX 3 with native video 🎬
FLUX 3 expands access from image generation into full multimodal content creation.

Berlin-based Black Forest Labs has shipped FLUX 3, a multimodal model that generates images and audio-video clips up to 20 seconds from a single prompt. The model accepts text, image, and video inputs and produces clips with native audio, multilingual dialogue, typography, and consistent character adherence. In blind evaluations, FLUX 3 outputs were favored over competitors including Runway, Kling, and Grok Imagine, putting competitive pressure on established video generation platforms. In a notable robotics collaboration, FLUX-mimic — built with Zurich-based mimic robotics — learns new factory tasks from roughly 30 minutes of demonstration data instead of the usual 30+ hours. The release marks BFL's bold expansion into embodied AI alongside multimodal content creation.

🏥 OpenAI rolls out ChatGPT Health for US users 🏥
Apple Health integration gives ChatGPT direct access to personal medical data.

OpenAI launched ChatGPT Health, a new feature allowing US adults to connect Apple Health data and supported medical records for more personalized, contextual health conversations. The integration enables ChatGPT to reference users' health metrics and history when answering wellness-related questions, marking OpenAI's most direct push into healthcare distribution. OpenAI stated that connected health data and related chats will not be used to train its foundation models or to target advertisements. Privacy safeguards and the explicit no-training policy are designed to address concerns about sensitive medical data. The feature is currently available to paid users in the United States, with broader rollout expected later.

🤖 Microsoft previews MAI-Image-2.5-Pro and Voice-2-Flash 🤖
Two proprietary models enter public rollout across Azure's AI services ecosystem.

Microsoft's superintelligence team unveiled two in-house models: MAI-Image-2.5-Pro for high-fidelity image generation and editing, and MAI-Voice-2-Flash for faster, lower-cost voice applications. Both have entered public preview and join Microsoft's expanding production lineup. MAI-Image-2.5-Pro targets creative and enterprise image workflows with improved generation quality, while MAI-Voice-2-Flash is optimized for latency-sensitive voice scenarios at reduced cost. The releases give enterprise customers first-party alternatives within Azure, reducing dependency on third-party model providers. This demonstrates Microsoft's commitment to building proprietary multimodal capabilities alongside its OpenAI partnership.

INTERESTING TO KNOW

🎙️ GPT-Live adds hands-free voice control to Codex 🎙️

OpenAI has integrated its GPT-Live audio model into the ChatGPT desktop app on macOS and Windows, enabling hands-free, full-duplex voice control for agentic coding tasks via Codex at conversational speed. A custom hotkey lets users invoke voice mode over any application, turning the desktop client into a voice-driven coding copilot for paid subscribers. Bloomberg reported the same voice technology may power OpenAI's upcoming dedicated home speaker hardware, signaling a push into consumer device distribution.

💰 Etched raises $300M at $10.3B valuation 💰

AI chip startup Etched has raised $300 million in new funding, pushing its total past $1 billion and its valuation to $10.3 billion. The company's novel architecture runs math blocks at half the usual voltage, boosting FLOPs density for transformer inference workloads at lower pricing than general-purpose GPUs. The rapid fundraising reflects strong investor appetite for alternatives to Nvidia's dominance, with Etched competing directly for data-center inference footprint against incumbent chipmakers.